Repository navigation
chore(agents): remove Hermes agent support and drop it from CI - #1523
Merged
Merged
Conversation
Hermes has become high-maintenance in CI — its image build fetches an install script from raw.githubusercontent.com, which frequently 429s on shared runners and fails the Docker Smoke Test — for a backend whose ROI no longer justifies that cost. Remove Hermes as a supported/built agent. Removed: - Dockerfile.hermes and docs/hermes.md - hermes target from Dockerfile.unified, Dockerfile.package, docker-bake.hcl - hermes from all CI: docker-smoke-test(.unified), pr-preview, build-images, pre-beta-build, snapshot-build, build-operator, nightly-build (variant strings + the nightly hermes) release-resolver case block). Also dropped the now-dead retry-curl assertions in docker-smoke-test.yml (the retried curl lived only in the removed hermes Dockerfile block). - hermes from charts/openab/values.yaml, config.toml.example, docs/config-reference.md, docs/image-tags.md agent list - hermes from README.md / README.zh-TW.md (CLI list, feature bullet, Other Agents table row, and the architecture diagram cell) - the Dockerfile.hermes comparison section in docs/grok.md Left intentionally: historical ADRs and code comments that cite Hermes as prior art, and CONTRIBUTING/steering references that point contributors to Hermes as a reference project — these are not agent-support wiring.
thepagent
approved these changes
Sep 4, 2026
chaodu-agent
added a commit
that referenced
this pull request
Sep 4, 2026
Brings in the Hermes removal (#1523) so this branch no longer builds or CI-tests the Hermes image — the flaky raw.githubusercontent.com 429 smoke failures were unrelated to this PR. No conflicts (ADC changes and the Hermes deletions touch disjoint regions of values.yaml / config-reference.md).
Collaborator
|
read https://eli5.openab.dev/howto-build-hermes-openab-image/ if you need to build your own hermes-oab images yourslf. |
4 tasks
10 of 11 tasks
thepagent
pushed a commit
to openabdev/openab-pty
that referenced
this pull request
Oct 1, 2026
openab removed Hermes (openabdev/openab#1523) and nightly-images dropped it in #28, which left the pre-beta path untouched; image.yml still built it on every PR and push, and it was the slowest job (7 min vs 2-3). Existing pre-beta-hermes / beta-hermes tags stay in GHCR but no longer update. Co-authored-by: chaodu-agent <274062505+chaodu-agent@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What this PR does
Removes Hermes as a supported/built agent and drops it from CI.
Why
Hermes has become high-maintenance in CI for diminishing return. Its image build fetches an install script from
raw.githubusercontent.com, which frequently returns 429 (rate limit) on shared CI runners and fails the Docker Smoke Test (both standard and unified). The retry hardening added earlier only masks it. Given the low usage/ROI, we're retiring the backend rather than keep paying that CI-flakiness cost.What was removed
Dockerfile.hermesanddocs/hermes.md(deleted)hermestarget fromDockerfile.unified,Dockerfile.package,docker-bake.hclhermesfrom all CI workflows:docker-smoke-test.yml,docker-smoke-test-unified.yml,pr-preview.yml,build-images.yml,pre-beta-build.yml,snapshot-build.yml,build-operator.yml, andnightly-build.yml(variant strings + the nightlyhermes)release-resolver case block + its comment)assert_occurrenceschecks indocker-smoke-test.yml— that retriedcurllived only in the removed hermes Dockerfile block, so the assertions no longer matched anythinghermesfromcharts/openab/values.yaml,config.toml.example,docs/config-reference.md, and the agent list indocs/image-tags.mdhermesfromREADME.md/README.zh-TW.md— the intro CLI list, the pluggable-backend feature bullet, the "Other Agents" table row, and the architecture-diagram cell (box alignment preserved)Dockerfile.hermescomparison section indocs/grok.mdIntentionally left in place
Historical ADRs and code comments that cite Hermes as prior art (e.g.
config.rs,openab-mcpbreaker/auth), plusCONTRIBUTING.md/ steering references that point contributors to Hermes as a reference project. These are not agent-support wiring — scrubbing them would be historical revisionism and out of scope.Verification
charts/openab/values.yamlparse as valid YAML.docker-bake.hclhas zerohermesreferences and thegrok→kimitarget boundary is intact.grok→mimocodetarget boundaries verified intact after block removal.hermesreferences remain (Dockerfiles, CI, chart, config examples, agent lists, README).helm-unittestCI covers it.Diffstat: 20 files, +14 / −441.