From 1fa4ae4c849d37e5e4f6b1e7c70336d9dbb26888 Mon Sep 17 00:00:00 2001 From: PhD Jonathan MERCIER Date: Sat, 14 Nov 2026 02:23:06 +0100 Subject: [PATCH] Publish verified develop projection (agentic-devlopment-cs7) Source: 64dd0245d21a1d6f3f3f10d31de9ad6f3731d442 Independent technical Judge: subagent-20261114_30; current maintainer approval relayed by coordinator. Generated-only inventory; no installed-host compatibility claim. --- distribution.json | 2 +- skills/cycle/SKILL.md | 33 ++- skills/cycle/references/candidate-identity.md | 16 +- skills/cycle/references/common-output.md | 31 ++- .../references/communication-contract.md | 11 +- .../cycle/references/requirements-output.md | 8 +- skills/cycle/references/session-evaluation.md | 70 ++++++ skills/cycle/references/ticket-authoring.md | 201 +++++++++++++++--- skills/implement/SKILL.md | 23 +- .../references/candidate-identity.md | 16 +- skills/implement/references/common-output.md | 31 ++- .../references/communication-contract.md | 11 +- .../references/requirements-output.md | 8 +- .../implement/references/ticket-authoring.md | 201 +++++++++++++++--- skills/iterate/SKILL.md | 2 +- .../iterate/references/candidate-identity.md | 16 +- skills/iterate/references/common-output.md | 31 ++- .../references/communication-contract.md | 11 +- .../iterate/references/requirements-output.md | 8 +- skills/iterate/references/ticket-authoring.md | 201 +++++++++++++++--- skills/judge/SKILL.md | 12 +- skills/judge/references/candidate-identity.md | 16 +- skills/judge/references/common-output.md | 31 ++- .../references/communication-contract.md | 11 +- .../judge/references/requirements-output.md | 8 +- skills/judge/references/ticket-authoring.md | 201 +++++++++++++++--- skills/plan/SKILL.md | 29 ++- skills/plan/references/candidate-identity.md | 16 +- skills/plan/references/common-output.md | 31 ++- .../plan/references/communication-contract.md | 11 +- skills/plan/references/requirements-output.md | 8 +- skills/plan/references/ticket-authoring.md | 201 +++++++++++++++--- skills/research/SKILL.md | 37 ++-- .../research/references/candidate-identity.md | 16 +- skills/research/references/common-output.md | 31 ++- .../references/communication-contract.md | 11 +- .../references/delegated-investigation.md | 5 + .../references/requirements-output.md | 8 +- .../research/references/ticket-authoring.md | 201 +++++++++++++++--- skills/test/SKILL.md | 14 +- skills/test/references/candidate-identity.md | 16 +- skills/test/references/common-output.md | 31 ++- .../test/references/communication-contract.md | 11 +- skills/test/references/requirements-output.md | 17 +- skills/test/references/test-value.md | 133 ++++++++++++ skills/test/references/ticket-authoring.md | 201 +++++++++++++++--- 46 files changed, 1875 insertions(+), 363 deletions(-) create mode 100644 skills/cycle/references/session-evaluation.md create mode 100644 skills/test/references/test-value.md diff --git a/distribution.json b/distribution.json index 0395fe6d..09f41910 100644 --- a/distribution.json +++ b/distribution.json @@ -1 +1 @@ -{"files":[{"mode":420,"path":".github/workflows/production-checks.yml","sha256":"1d684d8c8279cb0bfb4cfb41f441fc100a3bf32d7541d0543fe1e93002d60566"},{"mode":420,"path":"README.md","sha256":"471d29cb418f7403041a75d2bbe6f66f12cab461fd37466e8a62acd2d07104a5"},{"mode":420,"path":"hooks/hooks.json","sha256":"b6f20378f5a58bfe3a7d8218a03eb63ea7027ecb25e78eecf3cb6aebba28e484"},{"mode":420,"path":"plugin.json","sha256":"cb702d13789c3a058c0878371a734ce0d8841641d64f76eebd891050e05b92db"},{"mode":420,"path":"runtime.json","sha256":"e985a8bfd4ed66960d5afe2d72fa5a7648ee46b19b7a2776b4a382f49d2cf8cc"},{"mode":420,"path":"scripts/observe.pyz","sha256":"daeedb1183cb5e427dbed82f63d76beed152624f0477671a1cff0fdabdecb77b"},{"mode":420,"path":"skills/cycle/SKILL.md","sha256":"0482c588197c33fd755d7502d9f14a4c2441209a1d1bb736b6cb062b73338ae5"},{"mode":420,"path":"skills/cycle/assets/checkpoint-example.md","sha256":"bcfd2b08b793e5dd6df45b7c3fd42b476f365cb87163a20ee444dbbad3f94915"},{"mode":420,"path":"skills/cycle/assets/cycle-state.md","sha256":"3ac3bfa69962fb78f8aa98c002e17b0932b9311d39e8da9e1612cf5a79039121"},{"mode":420,"path":"skills/cycle/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/cycle/references/beads-initialization.md","sha256":"7f5231f2c24d56e29b3dc74baa105a5dc3c93ae673e9ab0ee65aedb759340379"},{"mode":420,"path":"skills/cycle/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/cycle/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/cycle/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/cycle/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/cycle/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/cycle/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/cycle/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/cycle/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/cycle/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"mode":420,"path":"skills/cycle/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/cycle/references/requirements-output.md","sha256":"013f7393f4625592ae87bf1019e16959e376b7a2dbafe7ae922e66b4e3c634de"},{"mode":420,"path":"skills/cycle/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/cycle/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/cycle/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/cycle/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/implement/SKILL.md","sha256":"b21f1f7af2462592f0171468f579f60b87ee7942c86c7043ef0d567db11127be"},{"mode":420,"path":"skills/implement/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/implement/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/implement/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/implement/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/implement/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/implement/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/implement/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/implement/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/implement/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/implement/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/implement/references/requirements-output.md","sha256":"3b6d14b68a977823abbc39b3bd4ad94e8708590d7d085c6b406d4f0d0a8a52bc"},{"mode":420,"path":"skills/implement/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/implement/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/implement/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/implement/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/iterate/SKILL.md","sha256":"ead0f261d1e508699b6acb3e7ac6711925b8f7cfe6b2a57b01fa1432224c93a1"},{"mode":420,"path":"skills/iterate/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/iterate/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/iterate/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/iterate/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/iterate/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/iterate/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/iterate/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/iterate/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/iterate/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/iterate/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/iterate/references/requirements-output.md","sha256":"cb9163883d419a0264c8da616da3701ef0e346a1a4275bc1e6b2cc70ecfd290e"},{"mode":420,"path":"skills/iterate/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/iterate/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/iterate/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/iterate/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/judge/SKILL.md","sha256":"8bb5dea976372b4b3e94d1b2969add47be5063d0a8439696fb1569ade6b8b2bc"},{"mode":420,"path":"skills/judge/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/judge/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/judge/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/judge/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/judge/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/judge/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/judge/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/judge/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/judge/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/judge/references/finding-verification.md","sha256":"df1a0ed4b400a8c68c4a984110b2746034eec8510cc6ec3b463a2a80469c0a40"},{"mode":420,"path":"skills/judge/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"mode":420,"path":"skills/judge/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/judge/references/requirements-output.md","sha256":"4d752aa667aefee81bca8d5cc15d11eb8262b730454b1a49bcebe1a7276c9335"},{"mode":420,"path":"skills/judge/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/judge/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/judge/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/judge/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/plan/SKILL.md","sha256":"9ee961d49e0a0423a2fa271ede160cf9493e3d108e8aeaec09b123f5e085bc9d"},{"mode":420,"path":"skills/plan/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/plan/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/plan/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/plan/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/plan/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/plan/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/plan/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/plan/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/plan/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/plan/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"mode":420,"path":"skills/plan/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/plan/references/requirements-output.md","sha256":"239535b2e6da860926439a10cd6955c3d40553868f116f752125349252af0c55"},{"mode":420,"path":"skills/plan/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/plan/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/plan/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/plan/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/research/SKILL.md","sha256":"b1a94a9e2c2dbe2b4ce8d04b9ff085c662998c4431b724f1c944f4ed20f0273a"},{"mode":420,"path":"skills/research/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/research/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/research/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/research/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/research/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/research/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/research/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/research/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/research/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/research/references/delegated-investigation.md","sha256":"9ddd5c903bb79e235a68abad4b927e196f6bd96243facab79ebfd0e8557f0e09"},{"mode":420,"path":"skills/research/references/knowledge-pointers.md","sha256":"f01a50a8276c35c41df24ed96e382d3de006de9e92ee60ab4873a5ef2637dd62"},{"mode":420,"path":"skills/research/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/research/references/requirements-output.md","sha256":"d26ddf7e4015ea1cfa59c8397e4e879fa63d5af282ca22d3cbc322b2b73cfb0f"},{"mode":420,"path":"skills/research/references/reuse-decision.md","sha256":"cca5c6b4c4118434d7623067263473919d5a749994760bb50547829d371ae51c"},{"mode":420,"path":"skills/research/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/research/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/research/references/state-of-the-art.md","sha256":"0c3103c118c1790106776d67887f243bf30d8c0b9203004a9fdb82c65cb8d34e"},{"mode":420,"path":"skills/research/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/research/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/test/SKILL.md","sha256":"8d15c323f1d074413cd26d7bbb9e98f4ae3b17ac4dcfad08220578a3b0134c19"},{"mode":420,"path":"skills/test/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/test/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/test/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/test/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"mode":420,"path":"skills/test/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/test/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/test/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/test/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"mode":420,"path":"skills/test/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"mode":420,"path":"skills/test/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/test/references/requirements-output.md","sha256":"25ac5ef77f8321f6d272e3a3ceae4fac7370ac027e4550db66a2e31b4436a282"},{"mode":420,"path":"skills/test/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/test/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/test/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"mode":420,"path":"skills/test/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"}],"format_version":1,"release_ready":false,"source_files":[{"path":"hooks/hooks.json","sha256":"364b3d1b81621ed45816160f7cb481cccd4250d4fe2a46bce9879c9985c7fd15"},{"path":"plugin.json","sha256":"aa18a1f498fd72391fc3cef4b01090ce3f11a311d798939666cd15b5b4be101c"},{"path":"scripts/observe.py","sha256":"8221f476aeae64783e81e9db503cb496d76891db0853b1b3a924eb05bb918359"},{"path":"skills/cycle/SKILL.md","sha256":"0482c588197c33fd755d7502d9f14a4c2441209a1d1bb736b6cb062b73338ae5"},{"path":"skills/cycle/assets/checkpoint-example.md","sha256":"bcfd2b08b793e5dd6df45b7c3fd42b476f365cb87163a20ee444dbbad3f94915"},{"path":"skills/cycle/assets/cycle-state.md","sha256":"3ac3bfa69962fb78f8aa98c002e17b0932b9311d39e8da9e1612cf5a79039121"},{"path":"skills/cycle/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/cycle/references/beads-initialization.md","sha256":"7f5231f2c24d56e29b3dc74baa105a5dc3c93ae673e9ab0ee65aedb759340379"},{"path":"skills/cycle/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/cycle/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/cycle/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/cycle/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/cycle/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/cycle/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/cycle/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/cycle/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/cycle/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"path":"skills/cycle/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/cycle/references/requirements-output.md","sha256":"013f7393f4625592ae87bf1019e16959e376b7a2dbafe7ae922e66b4e3c634de"},{"path":"skills/cycle/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/cycle/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/cycle/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/cycle/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/implement/SKILL.md","sha256":"b21f1f7af2462592f0171468f579f60b87ee7942c86c7043ef0d567db11127be"},{"path":"skills/implement/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/implement/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/implement/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/implement/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/implement/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/implement/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/implement/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/implement/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/implement/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/implement/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/implement/references/requirements-output.md","sha256":"3b6d14b68a977823abbc39b3bd4ad94e8708590d7d085c6b406d4f0d0a8a52bc"},{"path":"skills/implement/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/implement/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/implement/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/implement/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/iterate/SKILL.md","sha256":"ead0f261d1e508699b6acb3e7ac6711925b8f7cfe6b2a57b01fa1432224c93a1"},{"path":"skills/iterate/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/iterate/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/iterate/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/iterate/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/iterate/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/iterate/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/iterate/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/iterate/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/iterate/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/iterate/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/iterate/references/requirements-output.md","sha256":"cb9163883d419a0264c8da616da3701ef0e346a1a4275bc1e6b2cc70ecfd290e"},{"path":"skills/iterate/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/iterate/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/iterate/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/iterate/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/judge/SKILL.md","sha256":"8bb5dea976372b4b3e94d1b2969add47be5063d0a8439696fb1569ade6b8b2bc"},{"path":"skills/judge/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/judge/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/judge/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/judge/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/judge/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/judge/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/judge/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/judge/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/judge/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/judge/references/finding-verification.md","sha256":"df1a0ed4b400a8c68c4a984110b2746034eec8510cc6ec3b463a2a80469c0a40"},{"path":"skills/judge/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"path":"skills/judge/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/judge/references/requirements-output.md","sha256":"4d752aa667aefee81bca8d5cc15d11eb8262b730454b1a49bcebe1a7276c9335"},{"path":"skills/judge/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/judge/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/judge/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/judge/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/plan/SKILL.md","sha256":"9ee961d49e0a0423a2fa271ede160cf9493e3d108e8aeaec09b123f5e085bc9d"},{"path":"skills/plan/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/plan/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/plan/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/plan/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/plan/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/plan/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/plan/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/plan/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/plan/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/plan/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"path":"skills/plan/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/plan/references/requirements-output.md","sha256":"239535b2e6da860926439a10cd6955c3d40553868f116f752125349252af0c55"},{"path":"skills/plan/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/plan/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/plan/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/plan/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/research/SKILL.md","sha256":"b1a94a9e2c2dbe2b4ce8d04b9ff085c662998c4431b724f1c944f4ed20f0273a"},{"path":"skills/research/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/research/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/research/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/research/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/research/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/research/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/research/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/research/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/research/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/research/references/delegated-investigation.md","sha256":"9ddd5c903bb79e235a68abad4b927e196f6bd96243facab79ebfd0e8557f0e09"},{"path":"skills/research/references/knowledge-pointers.md","sha256":"f01a50a8276c35c41df24ed96e382d3de006de9e92ee60ab4873a5ef2637dd62"},{"path":"skills/research/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/research/references/requirements-output.md","sha256":"d26ddf7e4015ea1cfa59c8397e4e879fa63d5af282ca22d3cbc322b2b73cfb0f"},{"path":"skills/research/references/reuse-decision.md","sha256":"cca5c6b4c4118434d7623067263473919d5a749994760bb50547829d371ae51c"},{"path":"skills/research/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/research/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/research/references/state-of-the-art.md","sha256":"0c3103c118c1790106776d67887f243bf30d8c0b9203004a9fdb82c65cb8d34e"},{"path":"skills/research/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/research/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/test/SKILL.md","sha256":"8d15c323f1d074413cd26d7bbb9e98f4ae3b17ac4dcfad08220578a3b0134c19"},{"path":"skills/test/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/test/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/test/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/test/references/candidate-identity.md","sha256":"f5aec83b85a04e7f8d918b89ffcb0a14a571cc59c531a3112faa9fd412111fa3"},{"path":"skills/test/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/test/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/test/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/test/references/common-output.md","sha256":"9e3f72035f114b1506935c1a7fcb5e3bdc144bc9716a205e8ceb27c6ac790696"},{"path":"skills/test/references/communication-contract.md","sha256":"50483190a138fe1b9399e72aac02b7090e64b0b96b8bb151ed771f3d2705f707"},{"path":"skills/test/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/test/references/requirements-output.md","sha256":"25ac5ef77f8321f6d272e3a3ceae4fac7370ac027e4550db66a2e31b4436a282"},{"path":"skills/test/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/test/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/test/references/ticket-authoring.md","sha256":"4c333218b2e3b702f501185a07faadadc714ee0d4299c44b9db4f25c30b57fd1"},{"path":"skills/test/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"}],"source_head":"a87bf984a99be11d768b2b60cfce0a9530875949","source_state":"committed-source-candidate","source_tree":"d38384f00680848167f45afbee7ae7cb841c6bac"} +{"files":[{"mode":420,"path":".github/workflows/production-checks.yml","sha256":"1d684d8c8279cb0bfb4cfb41f441fc100a3bf32d7541d0543fe1e93002d60566"},{"mode":420,"path":"README.md","sha256":"471d29cb418f7403041a75d2bbe6f66f12cab461fd37466e8a62acd2d07104a5"},{"mode":420,"path":"hooks/hooks.json","sha256":"b6f20378f5a58bfe3a7d8218a03eb63ea7027ecb25e78eecf3cb6aebba28e484"},{"mode":420,"path":"plugin.json","sha256":"cb702d13789c3a058c0878371a734ce0d8841641d64f76eebd891050e05b92db"},{"mode":420,"path":"runtime.json","sha256":"e985a8bfd4ed66960d5afe2d72fa5a7648ee46b19b7a2776b4a382f49d2cf8cc"},{"mode":420,"path":"scripts/observe.pyz","sha256":"daeedb1183cb5e427dbed82f63d76beed152624f0477671a1cff0fdabdecb77b"},{"mode":420,"path":"skills/cycle/SKILL.md","sha256":"2342d58d92f24d3e02e4361918679a3375906c2458b3290f2a9c94007dae4da9"},{"mode":420,"path":"skills/cycle/assets/checkpoint-example.md","sha256":"bcfd2b08b793e5dd6df45b7c3fd42b476f365cb87163a20ee444dbbad3f94915"},{"mode":420,"path":"skills/cycle/assets/cycle-state.md","sha256":"3ac3bfa69962fb78f8aa98c002e17b0932b9311d39e8da9e1612cf5a79039121"},{"mode":420,"path":"skills/cycle/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/cycle/references/beads-initialization.md","sha256":"7f5231f2c24d56e29b3dc74baa105a5dc3c93ae673e9ab0ee65aedb759340379"},{"mode":420,"path":"skills/cycle/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/cycle/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/cycle/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/cycle/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/cycle/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/cycle/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/cycle/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/cycle/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/cycle/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"mode":420,"path":"skills/cycle/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/cycle/references/requirements-output.md","sha256":"68442648d5c19d1c92b7dce4a9ceea93ea7ac2e1a578761a68c3b3bd826f5c43"},{"mode":420,"path":"skills/cycle/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/cycle/references/session-evaluation.md","sha256":"6599db4bc886eeead1e02ccc423d431f2ce2cbb7001ca0b96158127a4ed757fc"},{"mode":420,"path":"skills/cycle/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/cycle/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/cycle/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/implement/SKILL.md","sha256":"5ef73cca3e18a10c69b2a86a147104f06a00c4300069c8588f5ff0b96715f36a"},{"mode":420,"path":"skills/implement/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/implement/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/implement/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/implement/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/implement/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/implement/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/implement/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/implement/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/implement/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/implement/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/implement/references/requirements-output.md","sha256":"18d0b0c91fbaf9d814de1cc80ec42e3b61dd35561dbdddd967ce5b0f527fe719"},{"mode":420,"path":"skills/implement/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/implement/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/implement/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/implement/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/iterate/SKILL.md","sha256":"f5e2be93d053974c210514cd9d7b3afe3f031613cefc55f009a6adb76a0bd63f"},{"mode":420,"path":"skills/iterate/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/iterate/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/iterate/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/iterate/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/iterate/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/iterate/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/iterate/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/iterate/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/iterate/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/iterate/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/iterate/references/requirements-output.md","sha256":"f06c16f9d22163876d7ab625cc9881833dacb78f0bd51eb37c6551d97414f25b"},{"mode":420,"path":"skills/iterate/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/iterate/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/iterate/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/iterate/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/judge/SKILL.md","sha256":"2bc152f563900c021f1321c98b5f57401d2e07aec81b8e534649639da8acb4e0"},{"mode":420,"path":"skills/judge/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/judge/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/judge/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/judge/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/judge/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/judge/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/judge/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/judge/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/judge/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/judge/references/finding-verification.md","sha256":"df1a0ed4b400a8c68c4a984110b2746034eec8510cc6ec3b463a2a80469c0a40"},{"mode":420,"path":"skills/judge/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"mode":420,"path":"skills/judge/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/judge/references/requirements-output.md","sha256":"0a4232befd3cf8ec9d4f5b04ff0d2f2d8a86d04573aed81a5cda265cee6e31c9"},{"mode":420,"path":"skills/judge/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/judge/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/judge/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/judge/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/plan/SKILL.md","sha256":"71aace7a9d92299955e5593095a1cc95769d5650e2d7c4e28a44b34e001fb2b1"},{"mode":420,"path":"skills/plan/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/plan/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/plan/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/plan/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/plan/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/plan/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/plan/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/plan/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/plan/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/plan/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"mode":420,"path":"skills/plan/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/plan/references/requirements-output.md","sha256":"1cfac2261ddf68652d3e7463d82b4c217fac94754a0ee6cde0d4f5076ab4c180"},{"mode":420,"path":"skills/plan/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/plan/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/plan/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/plan/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/research/SKILL.md","sha256":"50ce3dabbced58ceb02716f94acedc333f95c0d422d9756c332defe290bf806e"},{"mode":420,"path":"skills/research/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/research/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/research/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/research/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/research/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/research/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/research/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/research/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/research/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/research/references/delegated-investigation.md","sha256":"1b1fccfd0971b20690d7fe5572c5b5afc0380ee8c51e1c6b53c805e44b0554a1"},{"mode":420,"path":"skills/research/references/knowledge-pointers.md","sha256":"f01a50a8276c35c41df24ed96e382d3de006de9e92ee60ab4873a5ef2637dd62"},{"mode":420,"path":"skills/research/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/research/references/requirements-output.md","sha256":"8258a3589e62d2dae2195e2db4c3a03d1365835c3c74814abfd446522f2166bf"},{"mode":420,"path":"skills/research/references/reuse-decision.md","sha256":"cca5c6b4c4118434d7623067263473919d5a749994760bb50547829d371ae51c"},{"mode":420,"path":"skills/research/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/research/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/research/references/state-of-the-art.md","sha256":"0c3103c118c1790106776d67887f243bf30d8c0b9203004a9fdb82c65cb8d34e"},{"mode":420,"path":"skills/research/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/research/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"mode":420,"path":"skills/test/SKILL.md","sha256":"3a0605c283aa4e22befa6a20173c3c9e7dc70a03145a8ed66551dea5573d0184"},{"mode":420,"path":"skills/test/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"mode":420,"path":"skills/test/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"mode":420,"path":"skills/test/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"mode":420,"path":"skills/test/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"mode":420,"path":"skills/test/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"mode":420,"path":"skills/test/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"mode":420,"path":"skills/test/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"mode":420,"path":"skills/test/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"mode":420,"path":"skills/test/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"mode":420,"path":"skills/test/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"mode":420,"path":"skills/test/references/requirements-output.md","sha256":"dc58b4819a22c33beea3546b8cdbb92355b749c50b982499c0e18d438fc85bd3"},{"mode":420,"path":"skills/test/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"mode":420,"path":"skills/test/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"mode":420,"path":"skills/test/references/test-value.md","sha256":"4f56a3e5eb4e7f9a3ec6dcded860d5272c07ef77add900c6c1e1b09e1463b683"},{"mode":420,"path":"skills/test/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"mode":420,"path":"skills/test/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"}],"format_version":1,"release_ready":false,"source_files":[{"path":"hooks/hooks.json","sha256":"364b3d1b81621ed45816160f7cb481cccd4250d4fe2a46bce9879c9985c7fd15"},{"path":"plugin.json","sha256":"aa18a1f498fd72391fc3cef4b01090ce3f11a311d798939666cd15b5b4be101c"},{"path":"scripts/observe.py","sha256":"8221f476aeae64783e81e9db503cb496d76891db0853b1b3a924eb05bb918359"},{"path":"skills/cycle/SKILL.md","sha256":"2342d58d92f24d3e02e4361918679a3375906c2458b3290f2a9c94007dae4da9"},{"path":"skills/cycle/assets/checkpoint-example.md","sha256":"bcfd2b08b793e5dd6df45b7c3fd42b476f365cb87163a20ee444dbbad3f94915"},{"path":"skills/cycle/assets/cycle-state.md","sha256":"3ac3bfa69962fb78f8aa98c002e17b0932b9311d39e8da9e1612cf5a79039121"},{"path":"skills/cycle/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/cycle/references/beads-initialization.md","sha256":"7f5231f2c24d56e29b3dc74baa105a5dc3c93ae673e9ab0ee65aedb759340379"},{"path":"skills/cycle/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/cycle/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/cycle/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/cycle/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/cycle/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/cycle/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/cycle/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/cycle/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/cycle/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"path":"skills/cycle/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/cycle/references/requirements-output.md","sha256":"68442648d5c19d1c92b7dce4a9ceea93ea7ac2e1a578761a68c3b3bd826f5c43"},{"path":"skills/cycle/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/cycle/references/session-evaluation.md","sha256":"6599db4bc886eeead1e02ccc423d431f2ce2cbb7001ca0b96158127a4ed757fc"},{"path":"skills/cycle/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/cycle/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/cycle/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/implement/SKILL.md","sha256":"5ef73cca3e18a10c69b2a86a147104f06a00c4300069c8588f5ff0b96715f36a"},{"path":"skills/implement/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/implement/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/implement/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/implement/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/implement/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/implement/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/implement/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/implement/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/implement/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/implement/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/implement/references/requirements-output.md","sha256":"18d0b0c91fbaf9d814de1cc80ec42e3b61dd35561dbdddd967ce5b0f527fe719"},{"path":"skills/implement/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/implement/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/implement/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/implement/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/iterate/SKILL.md","sha256":"f5e2be93d053974c210514cd9d7b3afe3f031613cefc55f009a6adb76a0bd63f"},{"path":"skills/iterate/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/iterate/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/iterate/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/iterate/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/iterate/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/iterate/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/iterate/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/iterate/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/iterate/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/iterate/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/iterate/references/requirements-output.md","sha256":"f06c16f9d22163876d7ab625cc9881833dacb78f0bd51eb37c6551d97414f25b"},{"path":"skills/iterate/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/iterate/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/iterate/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/iterate/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/judge/SKILL.md","sha256":"2bc152f563900c021f1321c98b5f57401d2e07aec81b8e534649639da8acb4e0"},{"path":"skills/judge/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/judge/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/judge/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/judge/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/judge/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/judge/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/judge/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/judge/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/judge/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/judge/references/finding-verification.md","sha256":"df1a0ed4b400a8c68c4a984110b2746034eec8510cc6ec3b463a2a80469c0a40"},{"path":"skills/judge/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"path":"skills/judge/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/judge/references/requirements-output.md","sha256":"0a4232befd3cf8ec9d4f5b04ff0d2f2d8a86d04573aed81a5cda265cee6e31c9"},{"path":"skills/judge/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/judge/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/judge/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/judge/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/plan/SKILL.md","sha256":"71aace7a9d92299955e5593095a1cc95769d5650e2d7c4e28a44b34e001fb2b1"},{"path":"skills/plan/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/plan/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/plan/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/plan/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/plan/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/plan/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/plan/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/plan/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/plan/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/plan/references/judgment-rubric.md","sha256":"f0b1de03bcfe88f7eefb1f6687ff05c4553c3217b3c2dfa8cf110912901926a9"},{"path":"skills/plan/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/plan/references/requirements-output.md","sha256":"1cfac2261ddf68652d3e7463d82b4c217fac94754a0ee6cde0d4f5076ab4c180"},{"path":"skills/plan/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/plan/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/plan/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/plan/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/research/SKILL.md","sha256":"50ce3dabbced58ceb02716f94acedc333f95c0d422d9756c332defe290bf806e"},{"path":"skills/research/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/research/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/research/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/research/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/research/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/research/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/research/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/research/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/research/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/research/references/delegated-investigation.md","sha256":"1b1fccfd0971b20690d7fe5572c5b5afc0380ee8c51e1c6b53c805e44b0554a1"},{"path":"skills/research/references/knowledge-pointers.md","sha256":"f01a50a8276c35c41df24ed96e382d3de006de9e92ee60ab4873a5ef2637dd62"},{"path":"skills/research/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/research/references/requirements-output.md","sha256":"8258a3589e62d2dae2195e2db4c3a03d1365835c3c74814abfd446522f2166bf"},{"path":"skills/research/references/reuse-decision.md","sha256":"cca5c6b4c4118434d7623067263473919d5a749994760bb50547829d371ae51c"},{"path":"skills/research/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/research/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/research/references/state-of-the-art.md","sha256":"0c3103c118c1790106776d67887f243bf30d8c0b9203004a9fdb82c65cb8d34e"},{"path":"skills/research/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/research/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"},{"path":"skills/test/SKILL.md","sha256":"3a0605c283aa4e22befa6a20173c3c9e7dc70a03145a8ed66551dea5573d0184"},{"path":"skills/test/references/authorization-policy.md","sha256":"eb6a5f4566d6510a0f4675d110065b28c6ce2153f9195a11e891e5a1f26a8935"},{"path":"skills/test/references/beads-integration.md","sha256":"5701d38ccbc14dc9d84eed859402879fa7e2edb87188cde2aa2480703aca735a"},{"path":"skills/test/references/candidate-identity.example.json","sha256":"012c53acf9a7f38219f2c832056c9adb8bcfa607b81166b05357b752496b693f"},{"path":"skills/test/references/candidate-identity.md","sha256":"7086728944661c6ea8581b84629e9ce8f274d58647366272348d02cf363bbec3"},{"path":"skills/test/references/candidate-identity.schema.json","sha256":"737ebd21b6b816d9cd34a7a5d5304a48a4463a0774ce969c098fd4497f54a9a3"},{"path":"skills/test/references/candidate-scope.example.json","sha256":"8229fa2b23c13549d88742b81fb869f4a611342eb2b8fb16895b164977b24a32"},{"path":"skills/test/references/commit-report.md","sha256":"d0e7af9fc2dee198398fc1a92c5ab74d1509ee8e8b71e7f19e17e87fc7ed8bed"},{"path":"skills/test/references/common-output.md","sha256":"d4dc42d037e2bb069cc30e4be093b11f014d75e279501d4babb7d2242bc7179f"},{"path":"skills/test/references/communication-contract.md","sha256":"ed2f6371feafc9712b321ee6ad97a9755eee743e237931780ea1d79d3f610602"},{"path":"skills/test/references/requirements-baseline.md","sha256":"31d7f3d065a6f6573285f77004899d98d7a84b41fc1cacc1b0beaec4498e9cf1"},{"path":"skills/test/references/requirements-output.md","sha256":"dc58b4819a22c33beea3546b8cdbb92355b749c50b982499c0e18d438fc85bd3"},{"path":"skills/test/references/review-policy.md","sha256":"6ac19fd1cab7a50938daecb87fc6bd7fd115620710bd1a45a6be9fbc06e79932"},{"path":"skills/test/references/session-report.md","sha256":"5c20cbb2fbfee918d0345da5a13e24fbba1985e45ee351d2b3f1a0dff8d3f81a"},{"path":"skills/test/references/test-value.md","sha256":"4f56a3e5eb4e7f9a3ec6dcded860d5272c07ef77add900c6c1e1b09e1463b683"},{"path":"skills/test/references/ticket-authoring.md","sha256":"c181bfd564ba280018fca74fc2a2897068981c66b5135cacff7bcb06ac5be05d"},{"path":"skills/test/scripts/candidate_identity.py","sha256":"0826eba626f123fd78bdf3f18411d2a4f6437dbecc3a32cf8fc23fd3ce9b4c6b"}],"source_head":"64dd0245d21a1d6f3f3f10d31de9ad6f3731d442","source_state":"committed-source-candidate","source_tree":"0af7743b48f288be7f179d48e2b063725ce9b58b"} diff --git a/skills/cycle/SKILL.md b/skills/cycle/SKILL.md index 2d71507d..05574b81 100644 --- a/skills/cycle/SKILL.md +++ b/skills/cycle/SKILL.md @@ -55,6 +55,11 @@ a previous permission. This is an agent instruction, not a runtime cache. the first checkpoint or a Git/session transition; it governs reporting safeguards, not an up-front load of response templates. For a UI checkpoint apply [the checkpoint example](assets/checkpoint-example.md). + For an explicitly requested Skill evaluation in a source checkout, read + [evaluation piloting](references/session-evaluation.md) before launching runs. + Pilot the official Skill Creator CLI from the session; use its Goose runner + by default. Keep native delegation optional and evaluation separate from product acceptance. + Verify available filesystem, shell/test and delegation capabilities. Read the project's local instructions and inspect the repository without mutation first. Ask only for missing information that blocks safe progress; list nonblocking @@ -64,6 +69,9 @@ a previous permission. This is an agent instruction, not a runtime cache. Before creating or repairing task descriptions or acceptance criteria, apply [ticket authoring](references/ticket-authoring.md); verify readable purpose, criteria, ordered inputs and restart context without duplicating execution history. + Require explicit delivery mechanisms and asset maps for ready implementation work. + Route changed scope through consolidated ticket sections and reconciled dependency + edges; a populated tracker record alone does not prove execution readiness. If the repository uses Beads, apply [the Beads integration](references/beads-integration.md) before choosing work or mutating task state. Use fresh `bd prime` guidance; @@ -101,18 +109,23 @@ a previous permission. This is an agent instruction, not a runtime cache. when persisting or recovering a slice artifact or preparing a handoff, not for ordinary investigation. - - Identify the approved objective, roadmap source, milestone/epic/US and smallest - reviewable slice. If there is no roadmap, propose one and mark it unapproved. - - Inventory branch/status and pre-existing changes, capture candidate identity, - permissions and evidence. Report the starting checkpoint before mutations. + - Identify the approved objective, available roadmap/parent context and smallest + reviewable slice. Propose a missing roadmap only for requested planning or a + decision-blocking need; mark it unapproved. Otherwise report unavailable context; + do not expand a bounded investigation merely to fill hierarchy labels. + - Inventory branch/status, pre-existing changes, permissions and evidence. For + source-only observations record inspected revision/paths, dirty state and limits; + use preliminary or exact-candidate provenance only when the check's purpose + requires it. Report the starting checkpoint before mutations. - Keep required durable state in the agreed tracker or explicitly authorized workspace. Do not invent a default directory; verify the destination, conflicts and permission before writes. Do not persist private secrets, tokens, full sensitive tool outputs, or unnecessary personal data. If persistence is unavailable, emit a copyable handoff and disclose that no durable state was written. - Resume at the first incomplete or invalidated phase, not from the beginning. - Reinspect the working tree and compare candidate identity; treat old evidence - as stale if related files or environment changed. + Reinspect the working tree and select source-only observation, preliminary + check or exact-candidate proof under candidate identity; compare exact inputs + before proof reuse. Treat old evidence as stale if related files or environment changed. Treat roadmap outcomes, epic capabilities, user-story behavior and executable tasks as different levels. Keep distant work coarse and detail the ready frontier only. @@ -136,7 +149,11 @@ a previous permission. This is an agent instruction, not a runtime cache. safe actions without repeatedly asking. Request a real decision for unresolved trade-offs. A branch covers a US or slice, not necessarily a single task. - Use delegation only if available and useful. Give each worker the objective, + Use delegation only if available and useful, subject to required isolation or + independence. Coordinator baseline verification and bounded questions from + sufficient available sources need no worker; deeper Research investigations + follow that phase's dispatch policy. + Give each worker the objective, acceptance criteria, contract, approved scope, input artifacts and exclusive writable paths. Workers cannot coordinate implicitly. Parallel writers must never share a file. Workers return result, exact changes, evidence, limitations @@ -194,7 +211,7 @@ a previous permission. This is an agent instruction, not a runtime cache. - Verify the active objective/task, product-context coverage and freshness, current candidate, permissions and first incomplete or invalidated phase; retain the routing rationale and unmet prerequisites. - Prepare the immediate next action and blockers within current scope/permissions; collect handoff inputs, distinguishing recommendations from performed actions; format them after the output-contract application below. - - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. + - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Apply [the commit report](references/commit-report.md) only for a proposed, performed, failed or externally observed commit. Apply diff --git a/skills/cycle/references/candidate-identity.md b/skills/cycle/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/cycle/references/candidate-identity.md +++ b/skills/cycle/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/cycle/references/common-output.md b/skills/cycle/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/cycle/references/common-output.md +++ b/skills/cycle/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/cycle/references/communication-contract.md b/skills/cycle/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/cycle/references/communication-contract.md +++ b/skills/cycle/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/cycle/references/requirements-output.md b/skills/cycle/references/requirements-output.md index 4ac5aa08..4b8894fd 100644 --- a/skills/cycle/references/requirements-output.md +++ b/skills/cycle/references/requirements-output.md @@ -1,9 +1,9 @@ -# Cycle output and handoff contract v1.3.0 +# Cycle output and handoff contract v1.4.0 Apply this contract only to cycle. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### cycle @@ -15,8 +15,8 @@ and checked scope; parent/release gates remain separate. Optional filename: `cycle-state.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Phase delivery checklist diff --git a/skills/cycle/references/session-evaluation.md b/skills/cycle/references/session-evaluation.md new file mode 100644 index 00000000..cf24a79f --- /dev/null +++ b/skills/cycle/references/session-evaluation.md @@ -0,0 +1,70 @@ +# Pilot Skill evaluation + +Apply only to an explicitly requested Skill evaluation in a source checkout. + +## Official tool: default + +1. Load agent-plugins:skill-creator. Read its evaluation workflow and artifact contracts. +2. Resolve its installed executable or scripts/skill-creator.mjs launcher. Read --help. Do not install or copy a runtime. +3. Confirm the source root, Skill, canonical evals/evals.json and writable workspace. Keep one workspace per Skill/iteration. +4. Freeze source, scenarios and assertions. Give both variants equivalent fixtures and tool access; hide expected answers and assertions from candidates. Explanation cases grade final-output properties only. Canonical execution_assertions and navigation requirements need actual tool-event/order evidence; normalization may omit custom metadata, so preserve the canonical inventory separately. Never regrade historical outputs under changed criteria. +5. Confirm executor/grader models, baseline, run profile and spending bounds. Reuse verified choices; ask only for missing decisions. +6. Use without_skill to measure instruction contribution. Use old_skill plus a frozen baseline to compare versions. +7. Use fast for one exploratory pair per case; standard for three; release for five. Verify aggregate_budget for standard/release. Do not infer spending authority from it. +8. Verify supported provider transport without printing secrets. Do not extract keyring data, copy credential files or request secrets in chat. +9. Run validate, audit and design-evals. Resolve errors before model execution. Treat --dry-run as planning only. +10. Before model execution, verify one fresh context per case/variant and actual ambient instruction exclusion. A fresh cwd, --no-session or --no-profile alone does not establish exclusion of installed Skills, hints or ancestor instructions. Keep normal authentication unchanged; do not fake HOME or extract credentials to manufacture isolation. If exclusion is unverified, stop the clean-baseline experiment and report the missing host capability. Only after this gate run full-eval --execute from the current session's shell and let the official runner launch paired candidates. +11. Use subprocess grading by default. Let each separate blinded grader assess one output against all frozen semantic criteria; never self-grade or expose sibling outputs/prior grades. +12. Inspect the actual exit code, JSON status, phase results and recovery commands. Distinguish transport/host failures, incomplete evidence and behavioral failures. +13. Preserve failed workspaces. Fix only the authorized prerequisite; resume with --resume and unchanged bound inputs/settings. Use a new iteration after source/scenario changes. +14. If awaiting-grading, read the creator's delegated-grading workflow before importing grades. Do not substitute invented receipts or local adapter records. +15. Verify every required variant/repetition has outputs, timing or explicit unavailability, and bound grading. Capture duration at each invocation, not around a batch; record provider token/cost data only when actually exposed. Unknown metrics remain null with unavailable_reason; use a verified creator version supporting that local contract. Keep missing grades separate from present inconclusive grades, and inspect missing telemetry regardless of the primary status. Never infer acceptance or zero usage from exit 0. +16. Run aggregate and review --static if those phases remain incomplete. Use the official HTML viewer, not a custom readiness report. +17. Resume remaining official phases; run evaluation-profile verification and analyze. Inspect receipt completeness separately from pass/fail and human review. +18. Report completed/required runs, grades, failures, comparison limits and the HTML path. Keep review-ready distinct from behavioral failure; complete is not pass or release acceptance. +19. Stop on an unresolved transport, exhausted budget or missing authority. Return the exact prerequisite and next command; never repeat a known global failure across all suites. + +Run from the verified source root. Replace every placeholder before execution. +Set CREATOR to the verified installed source launcher; use its native executable instead when applicable. + +~~~sh +CREATOR=/absolute/agent-plugins/skills/skill-creator/scripts/skill-creator.mjs +SKILL=skills/cycle +NAME=cycle +WORKSPACE="evidence/runs//iteration-1" +MODEL="" +GRADER="" +node "$CREATOR" full-eval --help +node "$CREATOR" validate "$SKILL" +node "$CREATOR" audit "$SKILL" --format json +node "$CREATOR" design-evals "$SKILL/evals/evals.json" --skill-path "$SKILL" --format json +node "$CREATOR" full-eval "$SKILL" --workspace "$WORKSPACE" --execute --baseline without_skill --model "$MODEL" --grader review-a="$GRADER" --grading-mode subprocess --run-profile fast --concurrency 2 --provider-max-attempts 1 --max-grader-calls --format json +# Resume with the same execution arguments plus --resume; follow returned recovery commands. +node "$CREATOR" aggregate "$WORKSPACE" --skill-name "$NAME" +node "$CREATOR" review "$WORKSPACE" --skill-name "$NAME" --benchmark "$WORKSPACE/benchmark.json" --static "$WORKSPACE/review.html" +node "$CREATOR" verify "$SKILL" --profile evaluation --evaluation "$WORKSPACE" --format json +node "$CREATOR" analyze "$WORKSPACE" --skill-path "$SKILL" --format json +~~~ + +For a version comparison, replace --baseline without_skill with +--baseline old_skill --baseline-skill /absolute/frozen/skill. +Add approved provider cost/token/time caps supported by the installed help. +Do not widen retries, run counts or grading calls implicitly. + +## Native session adapter: optional + +1. Select this path only explicitly. Verify tools/session_eval.py in the source checkout; do not install it. +2. Use explain-level text cases and a new local workspace, never an official full-eval workspace. +3. Prepare requests. Delegate each request.json in a separate context with its supplied cwd, model and budget; hide criteria, sibling runs and parent conclusions. +4. Record only successful returned bytes with observed delegate IDs. Leave failed delegations pending. +5. Send each grading-request.json to a new independent grader. Import exactly one result per criterion with id, verdict, evidence_quote and rationale. +6. Run status; resume pending tasks without overwriting completed work. Generate HTML with the installed official viewer module. +7. Label results session-observed, not official full-eval receipts. Disclose ambient baseline exclusion, model/identity and telemetry limits; do not claim an OS sandbox or stable winner. + +~~~sh +python3 -B tools/session_eval.py prepare skills/cycle --eval-set skills/cycle/evals/evals.json --model --ids 25 +python3 -B tools/session_eval.py record --delegate-id --model +python3 -B tools/session_eval.py grade --delegate-id --model +python3 -B tools/session_eval.py status +python3 -B tools/session_eval.py report --viewer /runtime/dist/eval-viewer/generate_review.js +~~~ diff --git a/skills/cycle/references/ticket-authoring.md b/skills/cycle/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/cycle/references/ticket-authoring.md +++ b/skills/cycle/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria; diff --git a/skills/implement/SKILL.md b/skills/implement/SKILL.md index 4dfb17ef..c41e4596 100644 --- a/skills/implement/SKILL.md +++ b/skills/implement/SKILL.md @@ -71,6 +71,13 @@ a previous permission. This is an agent instruction, not a runtime cache. capabilities produce an explicit blocker, never simulated execution evidence. Use the coordinator's verified slice state when available; do not fabricate execution or acceptance. + Before implementation, verify that the ticket names its service/runtime mechanism, + ordered approach, inspected inputs, proposed outputs, preserved assets and exact + restart action. Missing delivery design is a readiness gap: repair it under + authorized planning scope or request its specific decision, not guess paths or + substitute generic safety prose. Compare native dependencies/owner with text and + verify current user authority instead of treating stale planning-only text as law. + Read the approved plan, acceptance criteria, local instructions and ownership ledger. Reinspect branch/status/diff and compare pre-existing hunks. Block conflicting ownership; do not overwrite another contributor's work. Recover missing context or return to Plan before edits. @@ -99,7 +106,19 @@ a previous permission. This is an agent instruction, not a runtime cache. Read relevant glossary/ADRs and preserve public contracts. Expected values come from specification, worked example or independent oracle, not copied production logic. Reuse suitable tests; finish one vertical behavior slice before the next rather than - writing all speculative tests then all code. + writing all speculative tests then all code. Before adding a test, name its product guarantee, + concrete incorrect behavior and marginal detection beyond existing checks; select + the cheapest valid oracle, not a test-count or coverage-percentage target. + Testability means a minimal meaningful public seam with controlled external inputs + and observable outputs/effects. Do not add speculative abstractions, exports of + private helpers, injection layers or production-only switches solely to multiply + tests. If the required seam is absent, report the specific gap and obtain scope for + the smallest justified change; do not bypass the required reproduction gate. + Within authorized rationalization, retain/merge/remove/replace tests with the + guarantee and discrimination evidence recorded; preserving every legacy test is + not required. Never delete a meaningful regression or weaken acceptance to get green. + Mock only dependencies: assert the subject's contract, exact known outcomes and + important errors, not the mock's setup or copied expected calculations. 3. [ ] **Implement minimal green and refactor** @@ -167,7 +186,7 @@ a previous permission. This is an agent instruction, not a runtime cache. Prepare the immediate next action and its exact prerequisites, owner when known and expected result; distinguish proposed from performed actions. Collect facts for the exact candidate, criterion IDs, reproduction/tests, change ledger and limits, and the inputs needed for a handoff; draft its state card and restart prompt only after reading the delivery guidance. - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. Combine overlapping reports without duplication. After reading delivery guidance, draft the response and, at an actual handoff, the fenced YAML state card and copy-ready restart prompt. + After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Combine overlapping reports without duplication. After reading delivery guidance, draft the response and, at an actual handoff, the fenced YAML state card and copy-ready restart prompt. 7. [ ] **Execute the handoff or stop** diff --git a/skills/implement/references/candidate-identity.md b/skills/implement/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/implement/references/candidate-identity.md +++ b/skills/implement/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/implement/references/common-output.md b/skills/implement/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/implement/references/common-output.md +++ b/skills/implement/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/implement/references/communication-contract.md b/skills/implement/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/implement/references/communication-contract.md +++ b/skills/implement/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/implement/references/requirements-output.md b/skills/implement/references/requirements-output.md index 4d1aa01e..a1d467ac 100644 --- a/skills/implement/references/requirements-output.md +++ b/skills/implement/references/requirements-output.md @@ -1,9 +1,9 @@ -# Implement output and handoff contract v1.3.0 +# Implement output and handoff contract v1.4.0 Apply this contract only to implement. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### implement @@ -16,8 +16,8 @@ per red/green loop; source-format checks do not prove agent behavior. Optional filename: `changes.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Phase delivery checklist diff --git a/skills/implement/references/ticket-authoring.md b/skills/implement/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/implement/references/ticket-authoring.md +++ b/skills/implement/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria; diff --git a/skills/iterate/SKILL.md b/skills/iterate/SKILL.md index e336a1d5..4d3035f7 100644 --- a/skills/iterate/SKILL.md +++ b/skills/iterate/SKILL.md @@ -115,7 +115,7 @@ a previous permission. This is an agent instruction, not a runtime cache. 6. [ ] **Read delivery guidance and draft the phase response** - After investigation and immediately before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. Before drafting a proposed, performed, failed or externally observed commit report, apply [the commit report](references/commit-report.md). Before drafting a planned session end, explicit wrap-up or recommended context transition report, apply [the session report](references/session-report.md). Combine overlapping reports without duplication. + After investigation and immediately before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Before drafting a proposed, performed, failed or externally observed commit report, apply [the commit report](references/commit-report.md). Before drafting a planned session end, explicit wrap-up or recommended context transition report, apply [the session report](references/session-report.md). Combine overlapping reports without duplication. After reading delivery guidance, draft the actual handoff fenced YAML state card and copy-ready restart prompt; distinguish proposed from performed actions. diff --git a/skills/iterate/references/candidate-identity.md b/skills/iterate/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/iterate/references/candidate-identity.md +++ b/skills/iterate/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/iterate/references/common-output.md b/skills/iterate/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/iterate/references/common-output.md +++ b/skills/iterate/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/iterate/references/communication-contract.md b/skills/iterate/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/iterate/references/communication-contract.md +++ b/skills/iterate/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/iterate/references/requirements-output.md b/skills/iterate/references/requirements-output.md index 139af08b..a0709c3b 100644 --- a/skills/iterate/references/requirements-output.md +++ b/skills/iterate/references/requirements-output.md @@ -1,9 +1,9 @@ -# Iterate output and handoff contract v1.3.0 +# Iterate output and handoff contract v1.4.0 Apply this contract only to iterate. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### iterate @@ -17,8 +17,8 @@ Capability-only waiting consumes zero attempts; no weaker criteria to make tests Optional filename: `iteration.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Phase delivery checklist diff --git a/skills/iterate/references/ticket-authoring.md b/skills/iterate/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/iterate/references/ticket-authoring.md +++ b/skills/iterate/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria; diff --git a/skills/judge/SKILL.md b/skills/judge/SKILL.md index 7b688bc2..6be94339 100644 --- a/skills/judge/SKILL.md +++ b/skills/judge/SKILL.md @@ -110,6 +110,16 @@ a previous permission. This is an agent instruction, not a runtime cache. Review each mandatory acceptance criterion, risk-based security/regression, maintainability and operability. Missing evidence is unknown, not zero defects. + For application-test evidence, challenge marginal detection value and independent + oracles: mock self-validation, copied expected calculations, vague assertions for + exact outcomes, success/error ambiguity and missing important error paths cannot + support the claimed guarantee. Review rationalization decisions against preserved + product coverage, not legacy volume; a green test count is not quality or acceptance. + Check observed failing counterexamples versus hypothetical faults and required + end-to-end CI wall time versus aggregate/local durations. Missing measurements or + discrimination stay explicit; do not infer CI acceptance from parallel local runs. + Keep this review risk-scoped; no exhaustive new test/mutation campaign or candidate + edits during Judge. Reproduce checks only when capabilities and permissions allow; do not mutate the candidate or repair code while judging. @@ -173,7 +183,7 @@ a previous permission. This is an agent instruction, not a runtime cache. 6. [ ] **Read delivery guidance and draft the phase response** - After investigation and immediately before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. Keep the canonical phase artifact within that report. Before drafting a proposed, performed, failed or externally observed commit report, apply [the commit report](references/commit-report.md). Before drafting a planned session end, explicit wrap-up or recommended context transition report, apply [the session report](references/session-report.md). Combine overlapping reports without duplication. + After investigation and immediately before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Keep the canonical phase artifact within that report. Before drafting a proposed, performed, failed or externally observed commit report, apply [the commit report](references/commit-report.md). Before drafting a planned session end, explicit wrap-up or recommended context transition report, apply [the session report](references/session-report.md). Combine overlapping reports without duplication. After reading delivery guidance, draft the actual handoff fenced YAML state card and copy-ready restart prompt; distinguish proposed from performed actions. diff --git a/skills/judge/references/candidate-identity.md b/skills/judge/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/judge/references/candidate-identity.md +++ b/skills/judge/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/judge/references/common-output.md b/skills/judge/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/judge/references/common-output.md +++ b/skills/judge/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/judge/references/communication-contract.md b/skills/judge/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/judge/references/communication-contract.md +++ b/skills/judge/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/judge/references/requirements-output.md b/skills/judge/references/requirements-output.md index 327b69a4..61144c95 100644 --- a/skills/judge/references/requirements-output.md +++ b/skills/judge/references/requirements-output.md @@ -1,9 +1,9 @@ -# Judge output and handoff contract v1.3.0 +# Judge output and handoff contract v1.4.0 Apply this contract only to judge. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### judge @@ -16,8 +16,8 @@ measurement/required independence stays explicit. No self-approval. Optional filename: `judgment.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Result block diff --git a/skills/judge/references/ticket-authoring.md b/skills/judge/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/judge/references/ticket-authoring.md +++ b/skills/judge/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria; diff --git a/skills/plan/SKILL.md b/skills/plan/SKILL.md index 2c353797..ff536394 100644 --- a/skills/plan/SKILL.md +++ b/skills/plan/SKILL.md @@ -112,6 +112,10 @@ a previous permission. This is an agent instruction, not a runtime cache. A task is small when its result can be reviewed and resumed independently, not because it edits few files. Do not split by function/file or phase mechanically. + Present the problem, expected user change, observable success and open decisions + before technical detail. Do not copy general work instructions into each ticket; + keep task-specific execution inputs below the product explanation and use native + tracker fields/shared-policy references for common metadata and procedures. For each task record: Goal; Parent outcome/specification; Scope and exclusions; Inputs and expected output; Observable acceptance/error paths; Blocking dependencies; Verification and public seams; Ownership and permissions; Durable context and restart. @@ -121,9 +125,22 @@ a previous permission. This is an agent instruction, not a runtime cache. to recover what, why, boundaries and checks. Fill gaps before declaring ready. Agree meaningful behavior seams before tests; reuse an already authorized seam - without repetitive confirmation. Record what each seam catches and misses. - - For the near-term implementation frontier, break work into coherent tasks with + without repetitive confirmation. For each product guarantee, identify a concrete + incorrect behavior, independent oracle and cheapest valid check; record what the + seam catches and misses. Plan distinct risk coverage, not one test per function, + file or implementation variation. Existing suite size is neither a preservation + obligation nor a target; authorized rationalization may retain, merge, remove or + replace checks without weakening acceptance. Prefer representative integration + journeys plus cheap direct rule cases; justify expensive repetitions by a distinct + risk/confidence question, budget and stop rule. Define required CI jobs and any + approved time/cost target separately from local summed test durations; unknown + end-to-end CI measurements remain unresolved, not inferred from parallel runs. + + For the near-term implementation frontier, name the delivery mechanism (service, + managed/self-hosted, runtime) and data/credential flow in the technical second + layer, then ordered steps and a verified Read/Create/Modify/Preserve/Exclude + file or asset map. Link common policies instead of repeating them; unknown + architecture blocks execution readiness. Break work into coherent tasks with exact target paths, exclusive ownership, prerequisites and expected outputs. For distant roadmap outcomes, record needs, risks and dependencies instead; do not invent file paths, numeric targets, owners or approval. Multiple tasks/commits may share one @@ -143,6 +160,10 @@ a previous permission. This is an agent instruction, not a runtime cache. Create or update issues only when requested/authorized in the agreed tracker. Inspect existing issues first, reuse matching work, verify supported fields and write goals, exclusions, criteria, dependencies, ownership and restart context. + On updates, consolidate current solution/steps/files rather than append competing + designs; move dated findings to notes, preserve criterion IDs/history, reconcile + native edges and check cycles. Readback must pass the human readability gate in + ticket-authoring.md, not merely confirm nonempty fields. Include Research's verified task-specific reading references: files or authorized local web captures, entrypoints/sections, reasons, order, source identity and freshness/accessibility limits. Do not copy full research into tickets or invent @@ -166,7 +187,7 @@ a previous permission. This is an agent instruction, not a runtime cache. - Verify the cited product-context baseline, requested planning level, scope and approval status; expose decision-critical gaps and carry nonblocking assumptions explicitly. - Check the planning artifact's criteria, verification matrix, review policy and next executable frontier; when tasks or tickets are in scope, verify inputs, outputs, dependencies, ownership, permissions and restart context. Read back actual tracker writes when applicable; distinguish proposed tickets from created ones. - Prepare the immediate next action and blockers; collect verified reading pointers and permissions for any proposed handoff; format them after the output-contract application below. - - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. + - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Apply [the commit report](references/commit-report.md) only for a proposed, performed, failed or externally observed commit. Apply diff --git a/skills/plan/references/candidate-identity.md b/skills/plan/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/plan/references/candidate-identity.md +++ b/skills/plan/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/plan/references/common-output.md b/skills/plan/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/plan/references/common-output.md +++ b/skills/plan/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/plan/references/communication-contract.md b/skills/plan/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/plan/references/communication-contract.md +++ b/skills/plan/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/plan/references/requirements-output.md b/skills/plan/references/requirements-output.md index 323fdc7f..2b1573ca 100644 --- a/skills/plan/references/requirements-output.md +++ b/skills/plan/references/requirements-output.md @@ -1,9 +1,9 @@ -# Plan output and handoff contract v1.3.0 +# Plan output and handoff contract v1.4.0 Apply this contract only to plan. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### plan @@ -17,8 +17,8 @@ to checks and capabilities before declaring ready. Optional filename: `plan.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Phase delivery checklist diff --git a/skills/plan/references/ticket-authoring.md b/skills/plan/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/plan/references/ticket-authoring.md +++ b/skills/plan/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria; diff --git a/skills/research/SKILL.md b/skills/research/SKILL.md index 81c715a5..7d6afbfa 100644 --- a/skills/research/SKILL.md +++ b/skills/research/SKILL.md @@ -47,7 +47,9 @@ a previous permission. This is an agent instruction, not a runtime cache. 1. [ ] **Frame the objective and verify authority** - Verify root and local instructions, requested outcome, available roadmap/task and - sources. Missing roadmap information is explicit, not permission to invent goals. + sources. Missing roadmap information is explicit; do not invent goals or expand + bounded research to fill hierarchy labels. Resolve decision-blocking context + only; roadmap proposals belong to requested planning or a decision-blocking need. - Before creating or repairing task descriptions or acceptance criteria, apply [ticket authoring](references/ticket-authoring.md); verify readable purpose, criteria, ordered inputs and restart context without duplicating execution history. @@ -72,12 +74,17 @@ a previous permission. This is an agent instruction, not a runtime cache. 2. [ ] **Select and dispatch bounded investigations** Establish the bounded product objective from verified inputs, then select the next - question. Delegate each selected investigation mode to a separate worker context. + question. The coordinator may directly verify a cited baseline's coverage/freshness + or answer a bounded question from sufficient available sources; record sources, + gaps and reuse rationale, not a fictional worker. Delegate deeper selected investigations + to separate worker contexts when discovery, traversal or comparison is needed + beyond those inputs. This direct path never waives mandatory isolation or independence. Before dispatch, read [the internal investigation contract](references/delegated-investigation.md). These are intent labels, not required tool names: workers map them to available tools. - Do not inject the full session or concatenate raw reports. If delegation is unavailable, - identify the blocker and obtain explicit permission for direct investigation; no silent - fallback. Select adaptively, not a mandatory four-worker batch. + Do not inject the full session or concatenate raw reports. If a deeper selected + investigation requires delegation and it is unavailable, identify the blocker and + obtain explicit permission for direct investigation; no silent fallback or waiver + of mandatory isolation/independence. Select adaptively, not a mandatory four-worker batch. | Mode | Question | Action and observable result | |---|---|---| @@ -104,9 +111,13 @@ a previous permission. This is an agent instruction, not a runtime cache. request is not research evidence. Missing facts remain unknown; ask when they block a coherent next decision. Read glossary/ADRs when domain terms or irreversible decisions matter. - When planning checks, freezing inputs or reusing evidence, apply - [candidate identity](references/candidate-identity.md). Include staged, untracked, - binary and relevant environment inputs; define output-only exclusions. Verify source - identity and freshness before accepting a worker's evidence as reusable. + [candidate identity](references/candidate-identity.md) and select the provenance path: + source-only observation records inspected revision/paths, dirty state and limits; + a preliminary check records command, inputs, environment and observed result with + limited provenance. Exact-candidate acceptance or proof reuse requires capture, + rerun and comparison under that contract, including staged, untracked, binary and + relevant environment inputs with output-only exclusions. Verify source identity + and freshness before accepting a worker's evidence as reusable. - Retrieve likely entry points, traverse callers/data flow, explore based on findings and verify hypotheses with concrete files, documentation or reproducible probes. Tie technical findings to the product outcome; cite paths, symbols and versions @@ -127,9 +138,11 @@ a previous permission. This is an agent instruction, not a runtime cache. 4. [ ] **Integrate evidence and prepare the decision** - - Verify actual worker invocation and completion status, source identity and result - locator for each selected mode; mark unneeded modes N/A with reason. Missing - delegation requires a blocker or explicit direct-investigation waiver. Deduplicate + - For delegated modes, verify actual worker invocation and completion status, + source identity and result locator. For direct work, record the coordinator's + sources, result and bounded-path rationale; mark unneeded modes N/A with reason. + Required unavailable delegation needs a blocker or explicit direct-investigation + waiver, never invented completion or waived independence. Deduplicate findings, resolve or expose conflicts, and mark insights observed, inferred or to-verify. Failed or partial results remain provisional. Compare the smallest viable alternatives, dependencies, risks and unknowns without premature design. @@ -156,7 +169,7 @@ a previous permission. This is an agent instruction, not a runtime cache. Verify product-context sufficiency, unresolved gaps and the immediate next action, owner and prerequisite; distinguish proposed from performed actions. Collect handoff inputs now; format them after the output-contract application below. - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. Before drafting a proposed, performed, failed or externally observed commit report, apply [the commit report](references/commit-report.md). Before drafting a planned session end, explicit wrap-up or recommended context transition report, apply [the session report](references/session-report.md). Combine overlapping reports without duplication. + After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Before drafting a proposed, performed, failed or externally observed commit report, apply [the commit report](references/commit-report.md). Before drafting a planned session end, explicit wrap-up or recommended context transition report, apply [the session report](references/session-report.md). Combine overlapping reports without duplication. After reading delivery guidance, draft the response and, at an actual handoff, the fenced YAML state card and copy-ready restart prompt. diff --git a/skills/research/references/candidate-identity.md b/skills/research/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/research/references/candidate-identity.md +++ b/skills/research/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/research/references/common-output.md b/skills/research/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/research/references/common-output.md +++ b/skills/research/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/research/references/communication-contract.md b/skills/research/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/research/references/communication-contract.md +++ b/skills/research/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/research/references/delegated-investigation.md b/skills/research/references/delegated-investigation.md index 9ffbd6b7..25d4499f 100644 --- a/skills/research/references/delegated-investigation.md +++ b/skills/research/references/delegated-investigation.md @@ -2,6 +2,11 @@ Read before dispatching a selected Research mode. Workers explore in separate contexts; the coordinator keeps product context and integrates compact evidence-bound results. This internal worker contract is not the user-facing five-section report or a command to launch every mode. +The coordinator can verify a cited baseline or resolve a bounded question from +sufficient available sources without dispatch. Record its provenance and gaps; +never report an uninvoked worker. Use this contract for deeper selected investigations +and any required independent/isolated work, not as a mandatory worker for every read. + ## Dispatch and ownership 1. Give each worker the product objective, question, mode, scope/exclusions, source identity, useful pointers, allowed probes, permissions, capabilities and explicit agreed turn/time/output budget. Do not inject the full conversation or private builder reasoning. Resolve missing budget/capability rather than assume an override took effect. diff --git a/skills/research/references/requirements-output.md b/skills/research/references/requirements-output.md index cea8fa23..26380d8d 100644 --- a/skills/research/references/requirements-output.md +++ b/skills/research/references/requirements-output.md @@ -1,9 +1,9 @@ -# Research output and handoff contract v1.3.0 +# Research output and handoff contract v1.4.0 Apply this contract only to research. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### research @@ -15,8 +15,8 @@ unresolved questions; Handoff. Separate observations from hypotheses and targets Optional filename: `research.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Phase delivery checklist diff --git a/skills/research/references/ticket-authoring.md b/skills/research/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/research/references/ticket-authoring.md +++ b/skills/research/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria; diff --git a/skills/test/SKILL.md b/skills/test/SKILL.md index 970092f3..238f2831 100644 --- a/skills/test/SKILL.md +++ b/skills/test/SKILL.md @@ -12,6 +12,7 @@ Select the verification intent; bind criteria, candidate and environment through | Intent / condition | Execute | Required result | |---|---|---| | Verify an implementation or regression repair | Bind criteria to the candidate, run available checks and record actual results | Reproducible criterion-to-check evidence, failures and coverage limits | +| Review test value or rationalization evidence | Inspect product guarantees, oracles, marginal detection and cost; propose scoped decisions without editing | Justified retain/merge/remove/replace matrix, detected faults and mandatory-CI measurement or explicit gaps | | Reuse previous proof after a change | Compare the proof's inputs/environment with current ones; rerun affected checks | Justified retained proof and new evidence for changed inputs | | Verify an installed package, real provider or rendered UI | Use the exact required artifact/environment and inspect actual behavior | Environment-specific observations, or explicit missing capability; mock proof is not substituted | @@ -87,9 +88,14 @@ a previous permission. This is an agent instruction, not a runtime cache. Read the project test instructions and verify commands, environment, tooling and side effects. Isolate data; production tests need explicit authorization. - Map mandatory criteria to checks: unit, integration, negative paths, - regression, build/install and real-provider behavior as required by scope. - Do not invent a command or assume a tool exists; inspect project scripts. + Map each product guarantee to a concrete incorrect behavior, independent oracle + and cheapest valid check; add tests for distinct risks, not each implementation detail. + Include unit, integration, negative paths, regression, build/install and real-provider + behavior only as required by scope. Do not invent commands; inspect project scripts. + When selecting/reviewing application tests or rationalizing a suite, apply + [test value and CI cost](references/test-value.md) before choosing checks or + proposing changes. Existing test volume is not a requirement to preserve it; + Test records decisions and evidence, never silently repairs the frozen suite. Deterministic checks must discriminate the requested symptom, not just show a command ran. Prefer public seams and independent expected values. For nondeterminism record @@ -137,7 +143,7 @@ a previous permission. This is an agent instruction, not a runtime cache. Prepare the immediate next action and exact prerequisites, owner when known and expected result. Preserve the candidate and artifacts; supply the criterion/check matrix, actual results, failures/repro steps, replay commands and limits, and the inputs needed for a handoff; draft its state card and restart prompt only after reading the delivery guidance. - After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and five-section report. Combine overlapping reports without duplication. After reading delivery guidance, draft the response and, at an actual handoff, the fenced YAML state card and copy-ready restart prompt. + After investigation and before drafting any user-facing phase response (including blocked or unchanged results), apply [the common output contract](references/common-output.md) and [this phase's delivery contract](references/requirements-output.md); apply their checklists and selected delivery form (compact when eligible, otherwise five-section report). Combine overlapping reports without duplication. After reading delivery guidance, draft the response and, at an actual handoff, the fenced YAML state card and copy-ready restart prompt. Announce the Judge context transition when applicable; Test reports evidence, not acceptance. diff --git a/skills/test/references/candidate-identity.md b/skills/test/references/candidate-identity.md index c3212e54..f6b7088b 100644 --- a/skills/test/references/candidate-identity.md +++ b/skills/test/references/candidate-identity.md @@ -1,9 +1,19 @@ -# Candidate identity v1.2.0 +# Candidate identity v1.3.0 Contract ID: agentic-development/candidate-identity -Contract version: 1.2.0 +Contract version: 1.3.0 -Use before planning checks, freezing a candidate or reusing evidence. Choose one path. +Use before planning checks, freezing a candidate or reusing evidence. Choose the +path matching the purpose; source reading alone is not an acceptance check. + +## Source-only observation + +Record inspected paths/sections, revision or file identity, current dirty state and +provenance limits. For bounded source reading, no capture CLI or full manifest is +required. Do not imply executed behavior, complete input coverage or exact-candidate +acceptance from a revision/hash. Refresh affected conclusions when inputs change. +Preliminary executed checks use A; exact-candidate acceptance or proof reuse uses B. +Reusing a cited observation with verified freshness does not turn it into B proof. ## A. Run a preliminary check diff --git a/skills/test/references/common-output.md b/skills/test/references/common-output.md index 9d37bb41..2c53d5d1 100644 --- a/skills/test/references/common-output.md +++ b/skills/test/references/common-output.md @@ -1,7 +1,7 @@ -# Common delivery output contract v1.3.0 +# Common delivery output contract v1.4.0 Contract ID: agentic-development/common-output -Contract version: 1.3.0 +Contract version: 1.4.0 Read after investigation and before drafting every user-facing phase delivery, including blocked, unchanged and standalone results. Reuse unchanged guidance @@ -9,7 +9,24 @@ already loaded in this context; verify it before delivery rather than rereading This document governs response formatting, not permission to act. Requirements must already have been resolved before the actions they govern. -## Mandatory five-section final report +## Compact bounded delivery + +For a bounded read-only or no-change phase result with no actual handoff, commit, +judgment or session-end boundary, a concise Markdown response may replace the full +five-section report. Retain goal and task (or unavailable identity), actual phase, +result, evidence and limits, preservation and current authority, and next action +with its prerequisite (or an explicit stop). State actual changes or their absence; +never omit a material blocker, failure, ownership conflict or unmet criterion. +A blocked result can use this path only when the same boundary conditions hold. +No cycle visual, empty headings or YAML card is required on this path. Do not turn +an ordinary answer into a handoff/session end merely to satisfy a template. + +Use the full five-section report for changed-source delivery, actual handoff, +commit, judgment or session-end boundaries, or an explicit full-report request. +An actual handoff retains the state card/restart prompt and boundary facts below; +compact delivery never waives provenance, permissions, review or tracker gates. + +## Mandatory five-section final report at full-delivery boundaries Keep these English labels and order; translate values into the user's language. The current-goal heading includes the actual phase, not a proposed future phase. @@ -43,7 +60,7 @@ proposed actions, and the next gate. Include applicable boundary facts below.> ## Report delivery and optional saving A phase report is generated output, not an assumed existing input. Deliver its -required content inline by default inside the five-section response; for longer +required content inline by default in the selected delivery form; for longer facts use concise evidence-bound summaries and verified accessible locators. Beads remains task truth; report filenames suggested by a phase are optional. @@ -69,7 +86,7 @@ and loop again belong in roadmap position, never in the step attribute. Done covers only the requested scope, not parent, release or deployment acceptance. A recommended transition does not execute or authorize it. -Show at least a compact cycle visual, using only evidence-backed state: +For the full report show at least a compact cycle visual, using only evidence-backed state: ~~~text Research [reused] → Plan [active] → Implement [pending] → Test [pending] → Judge [pending] @@ -79,7 +96,7 @@ Correction loop: Iterate → affected phase This example is hypothetical. Use active, verified, reused, pending, blocked or unknown as justified; loading a Skill does not verify a phase. Do not fabricate a roadmap, task, milestone or acceptance. If roadmap identity is unavailable, -say so and show known cycle progress only, with unknowns explicit. Keep distant +say so; in the full report show known cycle progress only, with unknowns explicit. Keep distant roadmap detail coarse; link verified task records instead of duplicating a tracker. Explain a nonstandard transition with its reason, reused/current proof and missing prerequisites. The next action names the immediate executable step, not a later @@ -108,7 +125,7 @@ A new artifact format is not permission to create files or a competing task data Read this contract after investigation and before drafting the response; before delivery verify it together with the current phase delivery contract. Reuse already loaded unchanged guidance rather than reload blindly. -- [ ] Keep the five final headings in order, with the actual step in current goal, including blocked/no-change work. +- [ ] Select compact delivery only under its conditions; otherwise keep the five final headings in order with the actual step in current goal. Blocked/no-change alone does not waive a full boundary report. - [ ] Identify the actual phase, verified task/status, goal and success criteria. - [ ] Separate observations from proposals, unknowns and hypothetical examples. - [ ] Report owned files separately from preserved user work; give review order and reasons. diff --git a/skills/test/references/communication-contract.md b/skills/test/references/communication-contract.md index eaf02e53..03448347 100644 --- a/skills/test/references/communication-contract.md +++ b/skills/test/references/communication-contract.md @@ -1,7 +1,7 @@ -# Communication contract v1.7.0 +# Communication contract v1.8.0 Contract ID: agentic-development/communication-contract -Contract version: 1.7.0 +Contract version: 1.8.0 Mandatory for every phase. This is an instruction contract, not a tool firewall. Respond in the user's language. Keep evidence and decisions inspectable; do not @@ -24,8 +24,9 @@ Use a full checkpoint at commit boundaries, judgment, session handoff, US closur session start, objective change, and slice completion. Do not repeat an unchanged full report after every tool call. Report discoveries that affect scope or safety immediately. When no files changed, say so; never invent a diff for a phase. -At phase delivery and handoff use the five-section final report in common-output.md; -this supersedes the six-field envelope for final messages, not its evidence safeguards. +At phase delivery use the form selected by common-output.md; actual handoffs use +the full five-section final report. +This supersedes the six-field envelope for final messages, not its evidence safeguards. ## Required structure and notification fields @@ -33,7 +34,7 @@ Read requirements-baseline.md before defining or changing requirements or verifi and before implementation, for Goals/Non-goals, FR/GWT and applicable quantified NFRs. After investigation and before drafting the phase response, read common-output.md and this phase's requirements-output.md, directly linked from SKILL.md, for the -five-section final report. Its roadmap position contains progress and the immediate +selected delivery form. In the full report roadmap position contains progress and the immediate next action; explanation carries evidence, limits and applicable checkpoint details. Notify actual files, branches/commits and session recommendations truthfully. Do not repeat unchanged detailed tables; use verified locators and concise summaries. diff --git a/skills/test/references/requirements-output.md b/skills/test/references/requirements-output.md index 018757d8..d5c35c3d 100644 --- a/skills/test/references/requirements-output.md +++ b/skills/test/references/requirements-output.md @@ -1,9 +1,9 @@ -# Test output and handoff contract v1.3.0 +# Test output and handoff contract v1.5.0 Apply this contract only to test. Read after investigation and before drafting the phase response, with common-output.md directly linked from SKILL.md. Check both before delivery. Operational requirements are separate in requirements-baseline.md. -The common five-heading Markdown envelope remains mandatory; this reference adds this phase's +The common contract selects compact delivery or the full five-heading Markdown envelope; this reference adds this phase's content, not another final report or permission to write artifacts. ### test @@ -11,13 +11,20 @@ content, not another final report or permission to write artifacts. Report content: verification report. Topics: Task and candidate; Goals and Non-goals; Acceptance-to-check matrix (FR/GWT and NFR IDs); Commands and environment; Results and artifact locators; Coverage and limitations; Failures/reproduction; -Judge handoff. For NFR record metric values/context or not-run/blocker; never +Judge handoff. For application-test selection/rationalization include guarantee, +incorrect behavior, independent oracle, marginal detection and retain/merge/remove/replace +decisions with actual/proposed status and lost/preserved coverage. Separate defects +actually detected (reproduction evidence) from hypothetical faults and unverified gaps. +Report required end-to-end CI wall time/environment/window separately from summed test +or aggregate compute durations; unavailable CI measurements stay UNKNOWN/NOT RUN. +Do not substitute test-count changes or parallel local timing for value or CI proof. +For NFR record metric values/context or not-run/blocker; never replace an approved target with a convenient observation. Optional filename: `evidence.md` — not a required input or automatic write. Deliver inline by default; save only through the common output contract’s explicit-destination, -permission and conflict checks. These topics belong inside explanation or its linked -report, not additional outer response headings. +permission and conflict checks. These topics belong in the selected delivery form or its linked +report; the full report places them in explanation, not additional outer headings. ## Phase delivery checklist diff --git a/skills/test/references/test-value.md b/skills/test/references/test-value.md new file mode 100644 index 00000000..96f0e630 --- /dev/null +++ b/skills/test/references/test-value.md @@ -0,0 +1,133 @@ +# Product guarantees and test value + +Apply when selecting or reviewing application tests, or analyzing a suite for +rationalization. This is not permission to edit tests or the frozen candidate. +Test diagnoses and proposes changes; authorized corrections go through Iterate +and Implement, then Test checks the changed candidate. + +## Contents + +- [Select evidence, not volume](#select-evidence-not-volume) +- [Challenge the oracle before counting a pass](#challenge-the-oracle-before-counting-a-pass) +- [Rationalize within authority](#rationalize-within-authority) +- [Measure the mandatory path honestly](#measure-the-mandatory-path-honestly) +- [Source and applicability](#source-and-applicability) + +## Select evidence, not volume + +1. Start from the approved product guarantees and important user journeys, not + existing test files or implementation structure. Include relevant errors, + cancellation, boundaries, security and observed incidents. Do not invent every + imaginable input combination. Missing expected behavior is a requirements gap. +2. For each selected test or coherent test family answer: + **Which incorrect observable product behavior does this detect that the other + selected checks do not already detect?** Record the guarantee/criterion ID, + concrete fault, independent oracle, public seam and evidence level. A distinct + environment or boundary can justify overlap; assertion/file count cannot. +3. Choose the cheapest valid check at that seam. Use direct deterministic cases + for business-rule variations; keep a representative integration for actual + wiring and a small set of operational journeys for required system guarantees. + Do not replace a required integration, installed-package, real-provider or UI + observation with a cheaper mock that cannot expose the same fault. +4. Add a case only for a distinct risk, boundary, incident or confidence question. + Existing tests are not a preservation obligation. There is not a fixed test-count ceiling + or minimum: a large suite can be justified, a small green suite can be worthless. + Do not target an arbitrary reduction percentage, coverage percentage or number. + +A compact selection matrix is enough; group genuinely equivalent cases rather +than produce an audit document per test. For a requested whole-suite audit, +account for every test/family; an uninspected subset stays unknown, not useless. + +| Guarantee / source | Incorrect behavior | Oracle / seam / level | Marginal detection value | Cost / required lane | +|---|---|---|---|---| +| Approved criterion ID | Concrete externally wrong result | Independent expected result, actual subject | Distinct fault or justified overlap | Measured or unknown; local/required CI/optional | + +## Challenge the oracle before counting a pass + +- Exercise the actual subject. A test that only calculates or asserts its own + setup is a tautology; copied production logic is not an independent oracle. + Use specification examples, known exact values or independently justified + properties. A test-only reimplementation of the subject proves nothing about it. +- A mock controls an external dependency, not the behavior being verified. + Returning the mock's value and asserting that same value is insufficient unless + a distinct product contract is discriminated (for example selecting the correct + customer's ID, forwarding a mandated parameter, or translating a dependency error). + Do not ban mocks or contractually meaningful interaction assertions wholesale. +- When the contract determines an exact outcome, assert it, including important + persisted effects or absence of forbidden effects. Non-null, truthy, nonnegative + or shape-only checks cannot substitute for the known result. Tolerances or + property assertions need a contract-based justification, not vague correctness. +- Specify one expected outcome per input/state. Do not accept success or error + interchangeably, catch failures to pass, silently skip missing prerequisites, + or branch on the observed result to choose whichever assertion passes. Assert + the success and important error paths separately with deterministic conditions. +- Check discrimination with a symptom-specific failing reproduction or a bounded + counterexample/mutation at the real seam when authorized: wrong exact value, + inverted decision, missing effect or swallowed error must make the relevant + test fail. Keep probes isolated, restore the candidate, and never mutate the + frozen candidate during Test. An existing pre-fix red can provide this evidence; + do not mandate an exhaustive mutation campaign for every test. Without execution, + label the fault hypothetical and discrimination unverified. Static red/green + establishes source conformance, not application or agent behavior. + +## Rationalize within authority + +For an authorized suite rationalization, inspect assertions, setup and observed +cost hotspots before optimizing scheduling. Profile slow families and repeated +campaign setup first, not just cheap individual assertions. If no timings exist, +record unknown cost and propose the smallest useful measurement. + +| Decision | Use when | Evidence needed | +|---|---|---| +| retain | Distinct product fault, security boundary, incident regression or required environment | Criterion, discriminating oracle and unique risk or justified overlap | +| merge | Checks duplicate the same guarantee or expensive setup | Representative integration plus cheap direct cases; preserve distinct faults | +| remove | Test validates only setup/internal detail or duplicates another check without distinct value | No unique product guarantee lost; identify retained coverage or explain irrelevance | +| replace | Guarantee matters but current test cannot expose its fault or is unnecessarily expensive | Better observable seam/oracle and relevant red-capable regression | + +Record test/family, decision, rationale, guarantee lost or preserved, replacement +or retained check, discrimination evidence, cost before/after (or UNKNOWN), and +actual or proposed status. Removal/fusion is allowed within approved scope; do not +weaken product acceptance or discard meaningful incidents to obtain green or faster +results. Missing required coverage remains open until a valid replacement passes. +Preserve unrelated work and keep the old evidence's candidate identity intact. + +Repeated costly end-to-end/provider/evaluation campaigns need a distinct risk or +confidence question, workload/sample justification, budget and stop rule. Prefer +one representative wiring journey and inexpensive rule cases when they expose +the same faults. Nondeterministic confidence requirements can justify repetition; +name the uncertainty being measured. Stop adding/running checks when selected +mandatory guarantees and scoped risk questions have adequate evidence, or stop +with explicit gaps when prerequisites/budget fail. Green count is not a stop rule. + +## Measure the mandatory path honestly + +Distinguish: +- individual and summed test durations: hotspots, often not end-to-end elapsed time; +- aggregate runner/compute time and costly service calls: total execution cost; +- required CI wall time: end-to-end mandatory workflow path, including checkout, + dependency setup, build, tests, artifacts and required job dependencies/joins. + +For CI claims identify the candidate, workflow/required jobs, runner environment, +parallelism/cache state, measurement start/end and observation window/runs. State +whether queue time is included. Compare only to an approved target and like-for-like +baseline; a local grouped run or summed timings do not prove the complete CI path. +Parallelism can reduce waiting without reducing aggregate work or improving test +value. Classify optional lanes explicitly; moving a required guarantee off the +mandatory path needs scoped approval, not an unreported speed optimization. +Unavailable CI measurement is UNKNOWN/NOT RUN, never an extrapolated pass. Targets +for downstream applications stay unresolved until their workload and approval are known. + +At delivery carry the selection/decision matrix, defects actually detected with +reproduction evidence versus hypothetical faults, retained guarantees, open gaps, +and mandatory-path wall time versus aggregate cost. Test count is secondary context, +not evidence of confidence, rationalization or acceptance. + +## Source and applicability + +Michael Bromley, [The Problem With Your AI Tests](https://www.michaelbromley.co.uk/blog/the-problem-with-your-ai-tests/), +7 September 2026 (consulted 28 October 2026), describes tautologies, weak assertions, +missing edge cases, conditional pass logic and implementation-mirroring setup. +Its large suite and reported review findings are the author's observations, not +benchmarks or efficacy evidence for this Skill. The selection, rationalization +and CI-cost procedure above also incorporates the user's disproportionate-suite +feedback; historical counts and timings are not portable acceptance targets. diff --git a/skills/test/references/ticket-authoring.md b/skills/test/references/ticket-authoring.md index 24bcccb9..a2d2cd87 100644 --- a/skills/test/references/ticket-authoring.md +++ b/skills/test/references/ticket-authoring.md @@ -1,7 +1,7 @@ -# Readable and restartable task tickets v1.1.0 +# Readable and restartable task tickets v1.3.0 Contract ID: agentic-development/ticket-authoring -Contract version: 1.1.0 +Contract version: 1.3.0 Apply before creating or repairing task descriptions or acceptance criteria in any agreed tracker; apply the Beads bridge when relevant. This is authoring guidance, @@ -14,10 +14,16 @@ not an enforcement engine or permission to mutate tickets. "Improvements" or an implementation detail that hides the user/problem result. 2. Put Why first: observed problem, affected user, impact and desired value. Use short sections/bullets, one idea per paragraph; separate facts from assumptions. + Explain the situation and what the reader will be able to do or understand + afterward, before naming architecture, storage or fixtures. Use plain language; + explain necessary domain terms. A commit or prior decision is context, not the problem. Do not paste operational preambles, transcripts or circular justification. + Do not invent an end user for technical/non-behavioral work: identify the + affected maintainer/system, concrete difficulty and verifiable deliverable. 3. Bound one independently reviewable result. Split when outcomes need different prerequisites or acceptance decisions, not by file or phase mechanically. - How describes the approach; acceptance criteria describe results, not + Technical detail comes after the problem, expected outcome, limits, success + criteria and open decisions. How describes the approach; acceptance criteria describe results, not implementation steps such as "edit three files". Detail only the ready frontier; do not invent paths, owners, estimates, metrics or approvals. 4. For bugs, state current behavior, expected behavior, minimal reproduction and @@ -30,38 +36,139 @@ not an enforcement engine or permission to mutate tickets. 6. Record blocking questions as Decision / owner / readiness impact, with the next resolution action. Mark unknown owners/inputs explicitly. Resolve decision-critical gaps before readiness; block the affected work, not unrelated permitted work. + Keep product decisions visible in their own section, not hidden in technical + vocabulary or presented as settled implementation details. State which work + can proceed and which depends on the answer; do not invent command names, + output formats or unavailable functionality to make a ticket look ready. 7. Store current specification in the description, canonical criteria in the acceptance field when available (otherwise in the description), results in notes/comments, and status/owner/prerequisite edges in native fields. Never maintain divergent criteria copies. Dated notes contain criterion IDs, candidate, checks/results, limits and next action; use locators, not full logs or secrets. + Do not copy general work instructions (authorization, review, Git or session + procedure) into every ticket. Reference the applicable shared policy instead; + retain only task-specific restrictions, review/evidence needs and exceptions. + Keep the first restart action specific to this task, not a generic work checklist. + Verify current facts against accessible sources before reusing technical starting + points; mark stale or unverified facts and their readiness impact. Removing a + promised feature or choosing a new storage contract is a scope change, not a wording repair. Preserve history; explain supersession in a dated note. Do not change criterion IDs, thresholds, scope, status or ownership merely to reformat; substantive changes need current scoped authority. Do not bulk-rewrite unrelated tickets or close parents. 8. Read back the rendered ticket and fields after authorized writes. Use the readiness checklist below. Readiness does not establish acceptance or closure. +## Concrete execution frontier + +For implementation work, explain the delivery mechanism in the technical second +layer: named service or technology, managed versus self-hosted, execution environment, +and what code creates/configures versus what an operator/runtime injects or executes. +Spell out necessary domain abbreviations. For credentials or data, show the +producer → transport → consumer path and bootstrap/recovery boundary without values. +"Inject externally" or "configure securely" is not a complete How. Unknown provider +capabilities need a decision/probe, not an invented implementation; use a small +text diagram only when it clarifies the flow. + +Detail ready implementation tickets, not every distant roadmap item. Before +calling an implementation ticket ready, provide a verified file/asset map: + +| Action | Required content | +|---|---| +| Read | Repository-relative existing path, relevant symbol/section, why, reading order | +| Create | Exact proposed path, responsibility, expected inputs/outputs; confirm no equivalent exists | +| Modify | Existing path and bounded change; identify shared-file coordination | +| Preserve | User changes, shared infrastructure/state, credentials and compatibility contracts | +| Exclude | Adjacent work owned by another ticket, with the relevant dependency | + +Existing inputs must be inspected; distinguish inspected paths from suggested or +unavailable references. Proposed files are not existing assets or automatically +approved architecture. Resolve architecture/ownership prerequisites before fixing +new paths; use a bounded decision/research ticket when they are unknown. Non-code +or provider-side tasks name the exact documents/resources/operations instead of +inventing source files to satisfy the table. A broad folder list and "read the +repository" are not a restart plan. + +Use ordered steps with an observable intermediate result and dependencies, not +an exhaustive script or generic phase boilerplate. The first restart action must +name its input, operation and expected result; "inspect inputs then implement" is +insufficient. Scope and estimated size follow one reviewable outcome; split when +bootstrap, runtime integration and production rollout require different inputs, +authority or acceptance. Do not duplicate all consumer implementations in a +foundation ticket. Choose task names that expose the delivered capability. + +## Consolidate updates, do not append competing specifications + +Read the current description, criteria and native dependency edges before editing. +When a decision or audit changes the design, rewrite the affected current sections +into one consistent specification. Put dated findings, previous decisions and +execution history in notes/comments or a linked evidence artifact. Do not append +successive "architecture update" blocks that contradict How, exclusions or criteria. +Preserve prior history and stable criterion IDs; explain supersession, deduplicate +checks and obtain authority for substantive scope/target changes. Never silently +weaken acceptance or archive away evidence still needed to prove it. + +Recompute prerequisite edges when scope changes. Keep native edges and text +consistent, reject cycles, and do not make architecture selection depend on the +implementation whose design it must select. Parent/epic hierarchy is not a blocker +edge. Record actual owner/claim in native fields; do not retain "owner unassigned" +boilerplate that contradicts the tracker. A new current user authorization may +supersede a ticket's historical planning-only grant for the bounded action; +verify and record it, without inferring deployment, rotation or Git permission. + +## Readability gate before writing and after readback + +A human unfamiliar with this conversation must be able to answer in a quick scan: +why, what replaces what, which service runs where, how data/credentials flow, +which files/assets change, what stays untouched, the first step and the success check. +Prefer a short summary, ordered implementation steps and a file table over repeated +policy paragraphs. Link common security/review contracts and specialize only the +risks of this task; never use a word limit to omit decision-critical facts. +Follow the project's content-language policy; otherwise author in the user's +language while keeping identifiers/API names exact. + +If these answers are missing, repair the ticket before implementation or clearly +mark the exact unresolved input and its decision owner. Do not declare a roadmap +item execution-ready merely because it has no native blockers. Readback checks +meaning and contradictions, not only nonempty fields or a passing JSON schema. + ## Description template Preserve the required facts; combine headings for small tasks. Omit non-applicable -optional detail with a reason; do not leave unexplained placeholders. +optional detail with a reason; do not leave unexplained placeholders. The product +questions come first; technical detail is a short second layer, not another work +protocol. Use native fields for owner, status, parent and prerequisite edges; when +unavailable, add a compact task-specific ownership/dependencies section below. ~~~markdown ## Why - + + -## What - - -## How - +## Expected outcome + + ## Scope and non-goals -- In: -- Out: +- In: +- Out: ## Acceptance criteria + + +## Open decisions + + + +## Technical starting point + + + + + +## File / asset map + + ## Reading references 1. check/seam and independent expected result> - - -## Ownership and dependencies - - resolution owner/action and readiness impact, or none> + ## Restart - - + + ~~~ +If criteria live in a separate acceptance field, read the description and that field +as one ticket; do not duplicate or hide them behind a technical report. Necessary +execution metadata remains recoverable through native fields and resolvable references, +not repeated policy text in the product explanation. + Keep references small and ordered. Separate source, product/specification docs and external references; explain each reading decision. Include canonical URLs for local captures; mark ephemeral/local-only availability and source freshness. @@ -102,34 +210,71 @@ unresolved or N/A with reason, never passed. ## Filled example -Hypothetical criterion, not execution evidence; other required ticket fields still apply. +Hypothetical ticket, not execution evidence. This example keeps one result narrow; +inline output and missing-destination behavior would need separate criteria if in scope. ~~~markdown Title: Preserve the previous report when its destination already exists -Why: A user must not lose an earlier report through an implicit overwrite. -What: Refuse an unapproved overwrite while keeping inline delivery available. -How: Check destination conflicts before saving; the implementation is not the criterion. -Acceptance: REPORT-1 in the acceptance field. +## Why +A user saving a new report to an occupied destination could lose an earlier report +without choosing to replace it. + +## Expected outcome +The user receives an explicit conflict message and keeps the earlier report intact. + +## Scope and non-goals +In: saving to an existing report destination without overwrite approval. +Out: approved replacement, renaming, inline output and missing destinations. + +## Acceptance criteria ### REPORT-1 — Preserve existing content Source: This hypothetical task's no-implicit-overwrite requirement. Given: The requested destination already contains an earlier report. -When: The agent attempts to save without explicit overwrite permission. +When: The user requests saving a new report without explicit overwrite permission. Then: The earlier report's bytes remain unchanged. Check: Compare the original bytes after the attempt; execution not-run. + +### REPORT-2 — Explain the conflict +Source: This hypothetical task's explicit-conflict-message requirement. +Given: The requested destination already contains an earlier report. +When: The user requests saving a new report without explicit overwrite permission. +Then: The response identifies the occupied destination as the reason saving was refused. +Check: Inspect the response for that destination and reason; execution not-run. + +## Open decisions +Exact conflict-message wording -> owner unknown -> decide before message-specific +assertions; REPORT-1 does not depend on that wording. + +## Technical starting point +Locate the existing report-saving entrypoint; its path is not yet verified. +A conflict check is a proposed approach, not the acceptance criterion. + +## Restart +Locate the saving entrypoint and a conflict-reproduction seam before implementation. +Native fields: owner unknown; no verified parent or prerequisite issue IDs. ~~~ -"Add a conflict-check function" belongs in How, not in Then. Cover inline delivery -and missing-destination behavior in separate criteria when they are in scope. +"Add a conflict-check function" belongs in technical detail, not in Then. Do not +turn an uncertain approach into a product promise or mix future scoring/replacement +capabilities into a read-only/preservation result. ## Readiness checklist - [ ] With only the ticket and its resolvable references, a new reader can explain - why the work matters, what is excluded, where to start and how success is observed. + the problem, expected change, how success is observed and what remains to decide + without technical vocabulary or generic work instructions obscuring them. +- [ ] Limits precede technical detail; the task-specific starting point is short, + current or explicitly unverified, and identifies where to start. - [ ] Title/outcome and task size are coherent; bug facts/reproduction are separated from hypotheses when applicable. - [ ] Criterion IDs/sources map to independent checks; relevant errors/boundaries and approved NFR definitions are present, without implementation-only acceptance. +- [ ] File/asset map separates inspected inputs, proposed outputs, modifications, + preserved work and exclusions; service/runtime and delivery flow are explicit + where relevant, with ordered steps or an exact design prerequisite. +- [ ] Current sections are consolidated, criteria/history preserved, native owner + and prerequisites consistent and cycle-free; no unblocked-but-undesigned readiness. - [ ] Required inputs resolve; owner, parent/prerequisite direction, permissions and blocking decisions are explicit. Missing mandatory facts block readiness. - [ ] Readback confirms fields/links, preserved history and unchanged criteria;