Document persistent-runner pattern for crq loop in agent harnesses by kristofferR · Pull Request #16 · kristofferR/coderabbit-queue

kristofferR · 2026-07-03T16:57:52Z

A single crq loop wait can exceed an hour when the queue is deep or the account is rate-limited. Agent harnesses commonly kill plain background shell jobs between turns, silently orphaning the wait — observed today when two consecutive background crq loop runs were killed mid-wait (~51 min and ~2 min in).

This documents the working pattern in SKILL.md and llms.txt:

Run crq loop under the harness's persistent long-running-task primitive (in Claude Code, Monitor with persistent: true), redirecting the findings JSON to a file and emitting a final CRQ_EXIT:<code> line as the completion event.
A killed runner loses nothing: the PR stays enqueued, and re-running the same crq loop re-attaches idempotently.
Don't substitute a hand-rolled crq status polling loop.

https://claude.ai/code/session_01BgW8cjUycHYKDZWaiRSLXY

Summary by CodeRabbit

Documentation
- Added an explicit guide for keeping crq loop alive in agent-based test harnesses that may stop background jobs between turns.
- Included an example setup using a persistent long-running task primitive and capturing results to a JSON file.
- Documented emitting a final CRQ_EXIT:<code> completion line, including why set +e may be necessary.
- Clarified that if the runner is killed, re-running the same crq loop command is idempotent and reattaches to the wait.

A crq loop wait can exceed an hour when the queue is deep or the account is rate-limited, and agent harnesses commonly kill plain background shell jobs between turns — the wait gets orphaned and looks like a hang. Document the pattern: run the loop under the harness's persistent long-running-task primitive (Claude Code: Monitor with persistent:true), redirect the findings JSON to a file, and emit a final CRQ_EXIT:<code> line as the completion event. Also note that a killed runner loses nothing — the PR stays enqueued and re-running crq loop re-attaches idempotently. Claude-Session: https://claude.ai/code/session_01BgW8cjUycHYKDZWaiRSLXY

coderabbitai · 2026-07-03T16:57:59Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: 3326ca5e-16a6-41d1-8459-90661c723dad

📥 Commits

Reviewing files that changed from the base of the PR and between 9c0cba7 and 8408260.

📒 Files selected for processing (1)

skills/coderabbit-queue/SKILL.md

📜 Recent review details

🔇 Additional comments (1)

skills/coderabbit-queue/SKILL.md (1)
70-96: LGTM!

📝 Walkthrough

Walkthrough

Documentation updates add guidance for running crq loop as a persistent task in agent harnesses, including set +e, exit-code emission, and safe reattachment after runner termination.

Changes

Persistent Loop Documentation

Layer / File(s)	Summary
Agent harness guidance for crq loop `llms.txt`, `skills/coderabbit-queue/SKILL.md`	Both files add sections describing persistent harness execution for `crq loop`, `set +e` handling, `CRQ_EXIT:<code>` emission, and re-running the same command after kill.
Estimated code review effort: 1 (Trivial)	~5 minutes

Possibly related issues

loop: survive process death — resumable wait state, ideally daemon-owned loop execution #12: The documentation update covers crq loop durability across harness termination and reattachment behavior.

Poem

A bunny keeps the loop in sight,
Through harness naps and nighttime flight.
With set +e, the exit sings,
And reattach brings back the springs.
🐇✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: documenting the persistent-runner pattern for `crq loop` in agent harnesses.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch docs/persistent-loop-runner

_{Comment @coderabbitai help to get the list of available commands.}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:da92bfe820

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".

The documented Monitor command relies on echoing crq loop's exit code as the completion event, but crq loop intentionally exits non-zero for actionable outcomes (10 = findings, 2 = timeout). Under an inherited set -e the shell dies before the echo and the event never arrives. Claude-Session: https://claude.ai/code/session_01W4RViwpNDBmT2fs3qhQNx7

kristofferR · 2026-07-03T20:56:19Z

@coderabbitai review

coderabbitai · 2026-07-03T20:56:24Z

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

kristofferR · 2026-07-03T21:54:41Z

@coderabbitai review

coderabbitai · 2026-07-03T21:54:46Z

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

coderabbitai

Actionable comments posted: 1

🤖 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.
Inline comments:
In `@skills/coderabbit-queue/SKILL.md`:
- Around line 78-84: The fenced Monitor({...}) example is missing a language
hint, which will trip markdownlint. Update the opening fence for this example to
include a language tag such as js or text, and keep the rest of the Monitor call
unchanged so the snippet remains valid and locatable.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: 708bc3cc-350d-4707-8da6-24fd17e37562

📥 Commits

Reviewing files that changed from the base of the PR and between 823a260 and 9c0cba7.

📒 Files selected for processing (2)

llms.txt
skills/coderabbit-queue/SKILL.md

📜 Review details

🧰 Additional context used

🪛 markdownlint-cli2 (0.22.1)

skills/coderabbit-queue/SKILL.md

[warning] 78-78: Fenced code blocks should have a language specified

(MD040, fenced-code-language)

🔇 Additional comments (1)

llms.txt (1)
79-87: LGTM!

Claude-Session: https://claude.ai/code/session_01W4RViwpNDBmT2fs3qhQNx7

kristofferR · 2026-07-04T08:29:22Z

@coderabbitai review

coderabbitai · 2026-07-04T08:29:43Z

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

kristofferR · 2026-07-04T09:02:22Z

@coderabbitai review

coderabbitai · 2026-07-04T09:02:27Z

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

chatgpt-codex-connectorBot reviewed Jul 3, 2026
View reviewed changes

Comment threadskills/coderabbit-queue/SKILL.md Outdated

coderabbitaiBot reviewed Jul 3, 2026
View reviewed changes

Comment threadskills/coderabbit-queue/SKILL.md Outdated

Add language tag to the Monitor example fence
8408260
Claude-Session: https://claude.ai/code/session_01W4RViwpNDBmT2fs3qhQNx7

kristofferR merged commit 4e81492 into mainJul 4, 2026
1 check passed

kristofferR deleted the docs/persistent-loop-runner branch July 4, 2026 11:13

kristofferR mentioned this pull request Jul 4, 2026
Count CodeRabbit's completion reply toward loop convergence #17
Merged

Navigation Menu

Document persistent-runner pattern for crq loop in agent harnesses - #16

Document persistent-runner pattern for crq loop in agent harnesses#16
kristofferR merged 3 commits into
mainfrom
docs/persistent-loop-runner

kristofferR commented Jul 3, 2026•
edited by coderabbitai Bot
Loading

Uh oh!

coderabbitaiBot commented Jul 3, 2026•
edited
Loading

Walkthrough

Changes

Uh oh!

chatgpt-codex-connectorBot left a comment

Uh oh!

Uh oh!

kristofferR commented Jul 3, 2026

Uh oh!

coderabbitaiBot commented Jul 3, 2026•
edited
Loading

Uh oh!

kristofferR commented Jul 3, 2026

Uh oh!

coderabbitaiBot commented Jul 3, 2026•
edited
Loading

Uh oh!

coderabbitaiBot left a comment

Uh oh!

Uh oh!

kristofferR commented Jul 4, 2026

Uh oh!

coderabbitaiBot commented Jul 4, 2026•
edited
Loading

Uh oh!

kristofferR commented Jul 4, 2026

Uh oh!

coderabbitaiBot commented Jul 4, 2026•
edited
Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

kristofferR commented Jul 3, 2026• edited by coderabbitai BotLoading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by CodeRabbit

Uh oh!

coderabbitaiBot commented Jul 3, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

kristofferR commented Jul 3, 2026

Uh oh!

coderabbitaiBot commented Jul 3, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kristofferR commented Jul 3, 2026

Uh oh!

coderabbitaiBot commented Jul 3, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitaiBot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

kristofferR commented Jul 4, 2026

Uh oh!

coderabbitaiBot commented Jul 4, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kristofferR commented Jul 4, 2026

Uh oh!

coderabbitaiBot commented Jul 4, 2026• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

kristofferR commented Jul 3, 2026•
edited by coderabbitai Bot
Loading

coderabbitaiBot commented Jul 3, 2026•
edited
Loading

coderabbitaiBot commented Jul 3, 2026•
edited
Loading

coderabbitaiBot commented Jul 3, 2026•
edited
Loading

coderabbitaiBot commented Jul 4, 2026•
edited
Loading

coderabbitaiBot commented Jul 4, 2026•
edited
Loading