fix(agents): cap issue-format optimizer retries - #2998
Merged
Conversation
stranske
added
agent:codex
stranske
added
agent:codex
|
Bugbot is not enabled for your account, so this pull request was not reviewed. Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs. |
📝 WalkthroughWalkthroughThe format-guard workflows now track trusted optimizer attempts, enforce a three-attempt limit, correlate accepted runs by issue title, and preserve attempt markers across ambiguous dispatch failures. Documentation, tests, and worker metadata were updated. ChangesFormat-guard retry control
Estimated code review effort: 4 (Complex) | ~45 minutes Sequence Diagram(s)sequenceDiagram
participant Issue
participant FormatGuardWorkflow
participant GitHubActions
participant IssueLabels
Issue->>FormatGuardWorkflow: unchanged non-conforming body
FormatGuardWorkflow->>Issue: inspect trusted attempt markers
FormatGuardWorkflow->>GitHubActions: dispatch format optimizer
GitHubActions-->>FormatGuardWorkflow: accepted run or ambiguous CLI result
FormatGuardWorkflow->>Issue: record numbered attempt marker
FormatGuardWorkflow->>IssueLabels: pause after third attempt
Possibly related issues
Possibly related PRs
Suggested labels: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
Owner
Author
|
Runner dispatch state for autofix on PR #2998. Do not edit. |
Contributor
🤖 Keepalive Loop StatusPR #2998 | Agent: Codex | Iteration 0/12 Current State
🔍 Failure Classification| Error type | infrastructure |
|
Contributor
Keepalive Work Log (click to expand)
|
stranske
temporarily deployed
to
agent-high-privilege
August 9, 2026 00:09 — with 
GitHub Actions
Inactive
GitHub Actions
Inactive
Contributor
|
Runner dispatch state for codex on PR #2998. Do not edit. |
stranske
temporarily deployed
to
agent-high-privilege
August 9, 2026 00:10 — with
GitHub Actions
Inactive
GitHub Actions
Inactive
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 864a97bd84
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".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Contributor
There was a problem hiding this comment.
Pull request overview
This PR introduces a sequential attempt cap to the agents-issue-format-guard routing logic to prevent unbounded re-dispatching to the issue optimizer, and updates workflow-text assertions to ensure the cap machinery is present in both the root workflow and the consumer template.
Changes:
- Add a format-guard attempt counter derived from prior
<!-- format-guard:... -->markers and stop routing once a fixed maximum is reached, applyingagents:auto-pilot-pause. - Emit a one-time “attempt cap” marker/comment when automated formatting is exhausted.
- Extend the existing workflow YAML text-based test to assert the attempt-cap logic is present and positioned before dispatch.
Reviewed changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| tests/workflows/test_agents_issue_optimizer_format_trigger.py | Adds assertions that the attempt-cap variables/marker/comment exist in both workflow copies and appear before the optimizer dispatch call. |
| templates/consumer-repo/.github/workflows/agents-issue-format-guard.yml | Adds attempt counting + cap behavior (pause label + exhaustion comment) to prevent unbounded sequential optimizer retries in consumer templates. |
| .github/workflows/agents-issue-format-guard.yml | Mirrors the same attempt counting + cap behavior in the root workflow copy. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
agents-workflows-bot
Bot
added
agent:needs-attention
agents-workflows-bot
Bot
added
agent:needs-attention
stranske
removed
needs-human
stranske-keepalive
Bot
added
agent:needs-attention
stranske
removed
needs-human
agents-workflows-bot
Bot
added
agent:needs-attention
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
.github/workflows/agents-issue-format-guard.yml (1)
303-309: 🩺 Stability & Availability | 🟠 Major | ⚡ Quick winIncrease and test the accepted-dispatch probe limit.
Line 306 limits the probe to 20 runs. During a burst, the accepted dispatch can fall outside this result set. The workflow then releases
agents:formatand can dispatch the optimizer again. Use the required limit of 100 and assert it in the workflow contract test.
.github/workflows/agents-issue-format-guard.yml#L303-L309: change--limit 20to--limit 100.templates/consumer-repo/.github/workflows/agents-issue-format-guard.yml#L303-L309: apply the same synchronized limit.tests/workflows/test_agents_issue_optimizer_format_trigger.py#L63-L87: assert that both workflows contain--limit 100.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In @.github/workflows/agents-issue-format-guard.yml around lines 303 - 309, Increase the accepted-dispatch probe limit from 20 to 100 in both workflow files, .github/workflows/agents-issue-format-guard.yml lines 303-309 and templates/consumer-repo/.github/workflows/agents-issue-format-guard.yml lines 303-309. Update tests/workflows/test_agents_issue_optimizer_format_trigger.py lines 63-87 to assert that both workflows contain --limit 100.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Outside diff comments:
In @.github/workflows/agents-issue-format-guard.yml:
- Around line 303-309: Increase the accepted-dispatch probe limit from 20 to 100
in both workflow files, .github/workflows/agents-issue-format-guard.yml lines
303-309 and
templates/consumer-repo/.github/workflows/agents-issue-format-guard.yml lines
303-309. Update tests/workflows/test_agents_issue_optimizer_format_trigger.py
lines 63-87 to assert that both workflows contain --limit 100.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: a93ca531-29b5-4420-9ee3-848dd7f3f3d7
📒 Files selected for processing (5)
.github/workflows/agents-issue-format-guard.ymldocs/ci/WORKFLOWS.mdlangsmith-fleet-worker-attempt.jsontemplates/consumer-repo/.github/workflows/agents-issue-format-guard.ymltests/workflows/test_agents_issue_optimizer_format_trigger.py
Contributor
Provider Comparison ReportProvider Summary
📋 Full Provider Details (click to expand)openai
anthropic
Agreement
DisagreementNo major disagreements detected. Unique Insights
🔍 LangSmith Traces |
Owner
Author
Closer disposition: verifier CONCERNS = false positive vs current mainAudited provider-comparison CONCERNS (comment at 2026-08-09T02:30:56Z) against live Claims vs current main
Disposition
Closer agent: cursor | round 2026-08-09T02:40:19Z |
24 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
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.
Closes #2991
Automated Status Summary
Scope
Loop safety in the
agents-issue-format-guard<->agents-issue-optimizerhandshake:templates/consumer-repo/.github/workflows/agents-issue-format-guard.ymland.github/workflows/agents-issue-optimizer.yml. The goal is that no single issue can be routed to the optimizer without bound, whatever the validator's verdict.Context for Agent
Related Issues/PRs
Tasks
templates/consumer-repo/.github/workflows/agents-issue-format-guard.yml:62-67, reading thestatefield already fetched at:59-60(for example[[ "$(jq -r '.state // "" | ascii_downcase' issue.json)" == "closed" ]]setsexempt=true), so no validation, comment, or dispatch occurs for closed issues.run-name:to.github/workflows/agents-issue-optimizer.ymlso every run carries the issue number regardless of trigger — e.g.Agents Issue Optimizer #${{ github.event.issue.number || inputs.issue_number }}— makingdisplayTitlecorrelatable for bothissuesandworkflow_dispatchruns.contains($issue)toendswith($issue)at.github/workflows/agents-issue-optimizer.yml:150andtemplates/consumer-repo/.github/workflows/agents-issue-format-guard.yml:245, so#4640is not counted against#464.--limit 100to thegh run listcall at.github/workflows/agents-issue-optimizer.yml:144-146, so the one-hour window is not truncated by the 20-run default.templates/consumer-repo/.github/workflows/agents-issue-format-guard.yml:121-269: count prior<!-- format-guard:markers on the issue and, at or above a fixed maximum, applyagents:auto-pilot-pause, post one comment recording that automated formatting is exhausted, and stop dispatching instead of routing again.Acceptance criteria
pytest tests/workflows/test_issue_optimizer_recursion_guard.py::test_recursion_guard_counts_workflow_dispatch_runspasses. The test feeds the guard'sjqfilter a fixture of fiveworkflow_dispatchrun records for one issue inside the window (the shape captured above), asserts the computed count is5so the-gt 3branch trips, and includes a#4640record that must NOT be counted against#464.contains($issue)in place ofendswith($issue)and remove therun-name:line sodisplayTitleis the bare workflow name, confirmtest_recursion_guard_counts_workflow_dispatch_runsFAILS with a computed count of0(the exact production symptom), then revert and confirm it passes again.pytest tests/workflows/test_issue_format_guard.py::test_closed_issue_is_exemptpasses: given anissue.jsonfixture with"state": "CLOSED", the resolve step setsexempt=trueand no dispatch occurs.pytest tests/workflows/test_issue_format_guard.py::test_routing_stops_at_attempt_cappasses: given an issue whose comments already carry the maximum number of<!-- format-guard:markers, the routing step appliesagents:auto-pilot-pauseand performs nogh workflow rundispatch.Summary by CodeRabbit
Bug Fixes
Documentation