[hhyykk]

Goal And Delivered Outcome

• Outcome basis / optional anchor: no written specification; demonstrated failures on current main are the basis.
• Goal/source and gap: Three checks fail on main and on unrelated PRs: the TypeScript content-digest single-owner test, the Codex App profile scheduler-hint test in tests/test_loopx_turn_driver.py, and the Frontstage Pages share-bundle smoke.
• Observable before → after, with the validation row that proves it: content_digest_single_owner.test.ts reported a second whole-value digest matcher at conversation_binding.ts; two scheduler_hint tests failed with KeyError; the bundle smoke required site/blog/context-delivery/example.py/index.html → all three pass with no lowered gate, budget or recorded exception. Proven by the Validation rows below.
• Issue/task and intended base: Self-contained request; base loopx-project/loopx:main. No issue closure claimed.

Failure	Introduced by	Root cause	Fix
Digest single owner	#7856 (111519af7)	Native recovery proof checks added a local /^[a-f0-9]{64}$/ regex	Reuse BARE_SHA256_PATTERN from content_digest.ts; register the file as a canonical consumer in the test inventory
scheduler_hint KeyError	ac0feb198 (default-on Codex context delivery)	With native delivery on by default, a turn-less quota should-run returns QUOTA_VALIDATION_FAILED (decision capture requires --turn-instance-id or --begin-turn), so the payload has no hint	The test inspects the raw scheduler contract, so it passes --context-delivery off; every scheduler assertion is unchanged, and the exit code is asserted before reading the hint
Frontstage bundle smoke	#7829 (07b555494)	The context-delivery post links a shipped example.py; the smoke classified .py as a page route	Classify .py as a static asset; the smoke still requires the file itself to exist

Author Declaration

• Written by: model_agent: Factory Droid (lead, review and commits) with OpenAI Codex CLI (implementation).

Implemented against

• Specification and revision: no written specification; behavior references are the branch diff against origin/main.
• Criteria:

Criterion (spec clause)	Disposition	Symbol / path	Test or command
Validate recovery proof digests through the shared digest owner	implemented	loopx/control_plane/collaboration/conversation_binding.ts	tests/control_plane_ts/content_digest_single_owner.test.ts; tests/control_plane_ts/conversation_binding.test.ts
Keep the scheduler-hint profile test on the raw contract after default-on delivery	implemented	tests/test_loopx_turn_driver.py	pytest tests/test_loopx_turn_driver.py
Resolve shipped .py references as assets in the share bundle	implemented	examples/frontstage-share-bundle-smoke.mjs	npm --prefix apps/presentation/dashboard run smoke:frontstage-share-bundle

• Self-check before submission: Reproduced each failure on main before the fix, read the full diff and ran the checks listed under Validation. Verified DCO trailers.

Scope And Continuation

• Completed scope and remaining work: Complete within this scope.
• Slice boundary / successor: N/A. The turn-less should-run validation failure under default-on delivery is documented behavior, not a regression. docs/reference/required-work-context.md requires machine consumers that parse should_run to request raw output with --context-delivery off. The heartbeat checklist always passes --turn-instance-id, and ordinary CLI calls without a native Codex declaration keep full delivery. The failure is explicit (QUOTA_VALIDATION_FAILED) with a fix-arguments recovery action.

Validation

Public-safe summaries only, including in HTML comments. Do not paste raw logs,
private data or prompts, private screenshots, credentials, internal URLs, connection
strings, or local paths. Use repository-relative commands without sensitive
arguments, aggregate results, and already-public CI links. If evidence is
private, report its category and limitations only; do not upload it to prove a claim.

• Tested revision: 84c6922
• Run state: finished
• Input classes: synthetic

Check kind	Result	Public-safe evidence / limitation
unit	passed	content_digest_single_owner.test.ts + conversation_binding.test.ts: 62 passed, 0 failed. Before the fix the digest test failed with the reported matcher.
unit	passed	tests/test_loopx_turn_driver.py + tests/test_chat_conversation_bindings.py: 236 passed, 0 failed. Before the fix the two profile tests failed with KeyError: 'scheduler_hint'.
integration	passed	smoke:frontstage-share-bundle and examples/blog-bilingual-index-smoke.mjs passed with LOOPX_SITE_RESOLVE_DESKTOP_RELEASE=0; the same offline mode reproduced the missing-file failure before the fix.
static	passed	loopx canary premerge --from-git-diff --git-diff-base origin/main: merge gate passed.

• Coverage and gaps: The release-resolving mode used by the workflow and the later Pages steps (docs builds, analytics, browser checks, deploy) were not run locally; CI on this PR covers them. Open PR docs(blog): clarify progressive context and post-release evidence #7852 also edits the context-delivery post but does not change the smoke and does not fix this failure.

See validation disclosure guidance.

Frontend / Visual Evidence

• UI impact: none.
• Before: N/A; no UI surface change.
• After: N/A; no UI surface change.
• First viewport and preview: N/A; no UI impact.
• States and viewports shown: N/A; no UI impact.
• Source data: none.
• Attention review: N/A; no layout change.
• Craft review: N/A; no visual change.

Type of Change

• Bug fix
• New feature
• Breaking change
• Refactoring (no functional changes)
• Documentation update
• Test update

LoopX Area

• Control plane (goals, todos, quota, scheduler, registry, runtime)
• Benchmark boundary (adapters, runners, verifiers, scoring, evidence)
• Capability or extension (providers, adapters, skills)
• Public docs or presentation surface (README, protocols, dashboard)
• Build, packaging, installer, or CI
• Host or runtime integration

Technical Direction

• Direction / acceptance reference, when applicable: N/A; restores existing checks on main.

Shared-authority RFC fixture impact

• Production-scale fixture schema: N/A; no migration or shared-authority qualification claimed.
• Semantic dimensions changed, or reviewed no-impact rationale: N/A; no authority routing or promotion change.
• Provider conformance arms run: N/A; no conformance claim.
• Read-only legacy/file/PostgreSQL three-arm rehearsal (required for promotion, runtime-routing, or compatibility-projection changes): N/A; none of these changes is proposed.

Boundary Checklist

• Reviewed diff and body for private state, credentials, raw traces, internal links and local machine paths; none included in this body. Test fixtures are synthetic.
• No maintainer-owned benchmark work duplicated.
• Change remains scoped to the request described above.
• No paired RFC changed.
• UI impact marked none.
• Every branch commit includes a DCO Signed-off-by trailer (git commit -s).

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-02T01:45:23.968532Z	84c6922	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[loopx-agent]

Reviewer: model_agent | gpt-6.1-sol | OpenAI | runtime_reported | reasoning_effort=max

动机 / Motivation

维护者和评审者执行主干验证，以及依赖源码恢复和静态文章链接的使用者。原有摘要单一 owner 检查、Codex App 原始调度契约测试和静态文章链接检查失败；修复后这三项在相同基线上有可区分的通过证据，原生恢复权限和缺失文件检查保留。

摘要 owner guard、两种 profile 拼写的原始契约测试和完整离线导出 smoke 的目标失败均已修复；原命令的旧 fake SDK 失败另行保留，不宣称全部主干或远端 CI 已绿。本次不改变机器设置或默认差分开关，不验收发布/部署、真实模型或安装版；不将作者测试、相关 PR 或 CI 状态当作本人的独立验收。

有价值的结果是恢复验证信号，让真实缺陷与既有夹具故障能被区分；不能靠删除断言、抬预算或改产品默认让检查看起来通过。原请求已经足够限定这批修复，无需新增用户配置或确认轮次。

改动思路 / Approach

用已有摘要 owner 取代重复表达式、让内部测试按已有 raw 契约取值，并继续检查真实静态文件，是比新增例外或改变产品默认更小的修复。本 PR 仅修三个可复现检查；原始假 SDK 失败和严格 CQR/合并门单独保留，相关 #7865 的重叠需要集成协调。

production 改动只把 native recovery 已有的裸 SHA-256 判断交回 content_digest.ts。[a-f0-9] 与 [0-9a-f] 字符集合相同，锚点和 flags 相同，source/thread/time/stop/recovery 条件均未移动。另两项是维护真实检查的输入和目标类型：内部机器消费者明确请求 raw 输出；.py 引用要检查静态文件，不应要求它是含 index.html 的页面。

没有新的持久状态、provider、权限、调度、机器设置或抽象层。直接复用原 owner 比添加 regex 例外、关掉完整检查或新增模式更小；做不到真实 file check 和原生负例，便不能仅凭本地统计批准。

具体改动 / Changes

完整范围 8a9e116283df2796926ea153241f48c9c8fc3dbb → 84c692296a8886817303cebf05c533eb03bc4dc5：4 文件，+8/-3，全部 diff 已读。没有覆盖这三项修复的独立综合书面规范，依据是本 PR 自包含请求和不可变基线的实际失败；另已读取基线 docs/reference/required-work-context.md 的 Machine default，其中明确要求解析 should_run 等字段的内部机器消费者使用 --context-delivery off。当前 PR body 在 review 期间补充了这一既有文档边界，head 未变；未把作者声明当作规范或验收。

关键代码讲解

• planContextRecovery 的两个 proof digest 使用现有 BARE_SHA256_PATTERN；content_digest_single_owner.test.ts 增加真实 consumer inventory，没有新增例外、缩扫描范围或改预算。
• test_quota_cli_codex_app_profile_is_explicit_and_applicable 加显式 raw presentation 参数，并把 exit-code 断言移到读取 hint 前。全部 scheduler 断言保留，机器默认与交互 Agent 的差分开关不变；带 native declaration 而缺 Turn 的拒绝仍是显式 QUOTA_VALIDATION_FAILED。
• assertRelativeReferenceExists 仅把 .py 放入既有 asset 分支。该分支仍检查文件真实存在，普通 page 分支仍要求 index.html。

相关 #7865 当前 4509737dc18b5756b467209cbe13ab0bf916399a 有完全相同的 digest owner/consumer 两处修改；我只读取关联契约和重叠，不额外审第三个 PR。其余 scheduler/bundle 修复在本 PR；集成时应协调这份重叠，不能据另一个 PR 的评审或测试替本 head 验收。

对主干的风险 / Risks

未发现当前 head 的阻断性代码问题。独立 native owner 比较包含两个 digest 字段的 20 个正反例、4 个权限/source/stop 拒绝和 5 个普通分支，base/head 输出完全相同，正向只允许 adopt_native_result，没有借此获得 dispatch/新 Turn 权限。

本地验证	结果和边界
当前 digest owner + conversation binding	PASS，62 项；不可变 base 的 single-owner guard 失败
原始 profile 两项	base FAIL：缺 scheduler_hint；当前 PASS，原断言保留
实际源码 CLI	PASS，6 次；两种 raw profile 拼写及 native 缺 Turn 拒绝，两端完整 hint/diagnostic 语义对照
完整离线 bundle smoke	base FAIL：错误要求 example.py/index.html；当前 PASS
实际 asset/page 检查	PASS，11 个观察：缺失/删除 .py 仍拒绝；两端导出的 example 内容相同
typecheck、Ruff、diff-selected premerge	PASS，5 direct +13 selected checks

源码 CLI 原输出保留了 44 个差异，来自独立 fixture 路径和压缩 host-facts 中的当前观察时间。对原 zlib/base64 chunks 解码，只归一化 fixture root 与相差两秒的 host_facts.generated_at 后，完整 hint、错误与恢复命令一致；没有丢掉字段、正文、flags、额度或 stale tolerance。原始差异和失败的未归一化 equality 断言仍保留。

原始完整 Python 命令没有通过，不能隐去。 相同命令在 base 为 167 PASS /37 FAIL /32 setup errors，当前为 169 PASS /35 FAIL /32 setup errors。当前剩余 67 个失败项的 case ID、错误类型和消息在 base 中完全相同；base 独有的两项恰是本次修好的 scheduler profile。普通/管家 fake Codex SDK 缺少真实要求的 modelProvider 确认，导致后续 receipt/HTTP 断言级联失败；有关测试、假 SDK 与产品 provider 路径五个文件在两端逐字相同。未把 69 与67 的完整失败集合冒称相等。

只在私有测试 hook 中让 synthetic SDK 回声实际请求的 provider，产品代码、权限、case 输入和断言均不改，当前同一 236 项全部通过。该运行覆盖实际 Python→Core/Node 和隔离 journal，但不把“兼容夹具通过”改写成“原命令全绿”。本地前端依赖最初缺失的两次 setup failure 也保留；准备同锁文件依赖后才取得上述 bundle base-red/head-green，未删弱断言。完整导出仅更换隔离输出路径、同一 source root 和绝对 import，避免固定临时目录互相覆盖；没有生产源码修改。

语义与 CI 对齐

现有 vocabulary、matcher owner、raw 消费契约和失败恢复义务被复用，没有引入通用 Goal 领域规则或把 enforced checks 称为 guidance。code review 可以依据当前不变量和已归因的旧失败通过；严格 change-quality 仍记录原 required 命令为 FAIL，并保留独立合并限制，没有使用 baseline waiver 或 admin bypass。远端 CI 未读取、轮询或等待。额外 native 交付夹具的 guard 已保存原 decision 后发生 transport failure，保留原记录且未重跑准入；不据此声称当前 Host 交付或采用资格。

我的整体评价 / Verdict

APPROVE，准确 head 84c692296a8886817303cebf05c533eb03bc4dc5，同时保留严格验证门的 hold。本次三处具体修复已由不可变 base 的实际失败、当前正向与负例验证；没有请求该 PR 修复无因果关系的 SDK 夹具问题，也没有把仍红的 required 原命令豁免为通过。合并前仍须由原验证 owner 按其契约处理该门并重新取得准确 head 的资格。

long_horizon 和 user_experience 的有界判断均为 preserved：原始恢复/停止/来源边界不变，正确的下载链接仍可用，用户没有新增设置或确认轮次。独立判断采用了“检查第一个实际有用结果和真实调用路径”的经验建议；没有继承历史 RFC 结论、奖励标签或模型效用结论。

用已有摘要 owner 取代重复表达式、让内部测试按已有 raw 契约取值，并继续检查真实静态文件，是比新增例外或改变产品默认更小的修复。本 PR 仅修三个可复现检查；原始假 SDK 失败和严格 CQR/合并门单独保留，相关 #7865 的重叠需要集成协调。没有页面布局变化，视觉检查不适用；release-resolving export、后续 Pages/deploy/browser 步骤、真实 native Host/model 采用仍未验证。已知兼容/ownership 需求得到保留，未新增兼容框架或强制无关迁移。没有合并。

English verdict: APPROVE at 84c692296a8886817303cebf05c533eb03bc4dc5 for the three bounded fixes. Native 62 tests, actual source/proof parity and full offline bundle checks pass. The unadapted Python command retains 67 independently attributed pre-existing fixture failures; strict change-quality and merge qualification remain on hold.