feat(agents): pre-PR build+test gate + tighter current-context lint#3023
Merged
feat(agents): pre-PR build+test gate + tighter current-context lint#3023
Conversation
…ontext lint
Three related changes to the triage/refresh pipeline:
1. triage-prompt.md — adds a "Pre-PR build + test gate" section
before the expert review. Runs `npm run precommit` (or equivalent
for doc-only diffs), capped at 2 build→fix iterations. Rationale:
expert review is expensive; don't spawn subagents on broken code.
And a red PR at push time makes the planned CI auto-fix loop
indistinguishable from drift-after-merge.
2. context-refresh-prompt.md — adds an explicit rule: bullet labels
must be factual nouns ("**Compliance storyboard remediation**"),
not editorial framing ("**Compliance storyboard gaps**"). The
lint now enforces this at the error level.
3. validate-agent-context.mjs — adds LABEL_BAN regex that hard-fails
when gap/risk/concern/narrative/stakeholder appears inside a
bold span. Tier and editorial are deliberately omitted (AdCP has
legitimate product-level uses: "Tier-2 Production Verified",
"editorial workflow"). Single-line regex — won't span paragraphs.
Sibling repos (adcp-client, adcp-client-python, adcp-go) get the
build+test-gate mirror in a follow-up.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
#2921 removed the hardcoded AdCP Agent Types section from server/src/addie/rules/knowledge.md (Addie now defers to search_docs for taxonomy questions). This assertion was pre-existing broken on main after #2921 merged; not specific to any PR. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This was referenced Apr 24, 2026
bokelley
added a commit
to adcontextprotocol/adcp-client
that referenced
this pull request
Apr 24, 2026
…#871) Mirrors adcontextprotocol/adcp#3023. Routine now runs the repo's build+test tier before spawning expert subagents, capped at 2 build→fix iterations. Prevents shipping known-broken drafts and keeps the auto-fix loop's "red PR = drift" signal clean. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
bokelley
added a commit
to adcontextprotocol/adcp-client-python
that referenced
this pull request
Apr 24, 2026
…#271) Mirrors adcontextprotocol/adcp#3023. Routine now runs the repo's build+test tier before spawning expert subagents, capped at 2 build→fix iterations. Prevents shipping known-broken drafts and keeps the auto-fix loop's "red PR = drift" signal clean. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
bokelley
added a commit
to adcontextprotocol/adcp-go
that referenced
this pull request
Apr 24, 2026
…#87) Mirrors adcontextprotocol/adcp#3023. Routine now runs the repo's build+test tier before spawning expert subagents, capped at 2 build→fix iterations. Prevents shipping known-broken drafts and keeps the auto-fix loop's "red PR = drift" signal clean. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
4 tasks
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
1 participant
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.
Summary
Closes the loop on tonight's bot-PR failures (#3006 shipped with unresolved `workos` reference, #3007 shipped with `vi.mock` hoisting bug — both would have been caught by running the build locally).
Pre-PR build+test gate in `triage-prompt.md` — routine now runs `npm run precommit` (or equivalent) before the expert-review step. Fails the PR if the build is red after 2 fix iterations. Expert review is expensive; don't spawn subagents on broken code.
Context-refresh rule in `context-refresh-prompt.md` — bullet labels must be factual nouns (`Compliance storyboard remediation`, not `...gaps`).
LABEL_BAN lint upgrade in `validate-agent-context.mjs` — hard-fails when `gap`/`risk`/`concern`/`narrative`/`stakeholder` appears inside a bold span. Tier and editorial are deliberately omitted (legitimate product terms: "Tier-2 Production Verified", "editorial workflow").
Sibling repos (adcp-client, adcp-client-python, adcp-go) get the build+test-gate mirror in a follow-up.
Follow-up not in this PR
CI auto-fix loop for `claude/*` branches when the base moves after push. Brian flagged this — "things change underneath." Deferred because it needs a new routine or event-type handling, and tonight's gate already addresses push-time correctness.
Test plan
🤖 Generated with Claude Code