Skip to content

refactor: reduce wait-for-action to one PR monitor - #41

Merged
archae0pteryx merged 3 commits into
mainfrom
refactor/single-pr-monitor
Nov 15, 2026
Merged

archae0pteryx merged 3 commits into
mainfrom
refactor/single-pr-monitor

Conversation

@archae0pteryx

@archae0pteryx archae0pteryx commented Nov 15, 2026 •

Copy link
Copy Markdown
Contributor

Summary

  • Replace the multi-mode dispatcher with one read-only Python PR monitor.
  • Delete reviewer-specific watchers, triggers, workflow-only waits, tuning flags, and obsolete tests.
  • Retain head tracking, bounded requests, distinct check suites, superseded-run handling, and unconfirmed reviewer reporting.
  • Update dependent skills, README, agent metadata, and generated catalogs.
  • Use private system temporary directories and flatten remote text before printing status lines.

Interface

python3 skills/wait-for-action/scripts/wait_for_action.py [--pr N] [--repo OWNER/REPO] [--ttl 15m] [--log-dir PATH]

Five-second polling and fail-fast are fixed. No compatibility modes or triggers remain.

Validation

  • 112 tests pass.
  • Generated manifests and whitespace checks pass.
  • Two fresh-context Code Reviewer passes completed. Actionable caller and output-safety findings addressed; final follow-up has regression tests, not a fresh no-findings verdict.
  • Live read-only PR monitoring exercised successfully.

Review disposition

  • Fixed CI-failure/review-state guidance, stale reviewer descriptions, generic thread reporting, script path resolution, reviewer independence checks, private artifacts, and status-line forgery.
  • Documented API call volume without reintroducing polling modes or rate-limit waiting loops.
  • Retained standard argparse usage errors and documented Python 3.10+ runtime rather than adding wrappers or runtime managers.
  • Version changes remain part of the separate release process.

Summary by CodeRabbit

  • Documentation
    • Updated review guidance to observe existing CI and bot-review activity without triggering reviews, with a local reviewer used when no independent verdict is available.
    • Reworked the activity-wait guide around a single read-only monitor, including its options, behavior, and outcomes.
    • Removed instructions for targeted reviewer, check, and workflow waits.
  • Bug Fixes
    • The monitor now reports activity at most every five seconds, stops immediately on reported failures, and sanitizes displayed text.

@coderabbitai

coderabbitai Bot commented Nov 15, 2026 •

Copy link
Copy Markdown

Explain Change →

No actionable comments were generated in the recent review.

Walkthrough

The PR replaces specialized review and shell waiters with a Python tool that observes pull-request activity without triggering reviews. It updates the independent-review workflow, usage documentation, integrations, and tests to describe and use the new monitor.

Changes: PR activity monitoring

Layer / File(s) Summary
Python waiter and CLI
skills/wait-for-action/scripts/wait_for_action.py, skills/wait-for-action/scripts/coderabbit.py, skills/wait-for-action/scripts/pr_agent.py, skills/wait-for-action/scripts/wait-for-action.sh, tests/test_pr_activity.py, tests/test_coderabbit.py, tests/test_pr_agent.py, tests/test_wait_for_checks.py, tests/test_wait_for_ready.py
The Python waiter adds duration parsing, sanitizes emitted text, always prints activity, fails immediately on reported failures, and polls at intervals of up to five seconds. It removes the shell waiter and reviewer-specific scripts. Tests now exercise the Python CLI and its supported options.
Independent-review workflow
skills/independent-review/SKILL.md, skills/independent-review/.cursor-plugin/plugin.json, skills/independent-review/agents/openai.yaml, .claude-plugin/marketplace.json, .cursor-plugin/marketplace.json, .github/plugin/marketplace.json, README.md
The workflow uses one activity wait, evaluates current-head verdicts, and checks reviews again after pushes when reviewers had findings. Its descriptions specify that it does not trigger bot reviews and uses a fresh-context local reviewer when no independent verdict is available.
Waiter integration and usage
skills/lets-finish/SKILL.md, skills/wait-for-action/SKILL.md, skills/wait-for-action/.cursor-plugin/plugin.json, README.md, .claude-plugin/marketplace.json, .cursor-plugin/marketplace.json, .github/plugin/marketplace.json
The lets-finish instructions invoke the Python monitor and separately inspect CI results. The skill, README, and plugin descriptions document the monitor’s command, options, observation behavior, results, and artifacts.

Estimated code review effort: 4 (Complex) | ~45 minutes

Change: Refactor

Sequence Diagram(s)

sequenceDiagram
  participant IndependentReview as independent-review
  participant Waiter as wait_for_action.py
  participant GitHub as GitHub PR API
  participant LocalReviewer as Local reviewer
  IndependentReview->>Waiter: Invoke read-only activity wait
  Waiter->>GitHub: Observe PR checks, statuses, reviews, and comments
  GitHub-->>Waiter: Return observable PR activity
  Waiter-->>IndependentReview: Provide saved activity and result
  IndependentReview->>IndependentReview: Evaluate current-head verdicts and findings
  IndependentReview->>LocalReviewer: Request fresh-context review when no independent verdict is available
Loading
Merge Risk: Minimal · up to ba235

This PR consolidates the wait-for-action tooling into a single read-only PR monitor and updates the dependent skills and docs to match. No concrete merge-blocking risk was identified in the supplied changes.

Pre-merge checks: Passed 4

Check name Status Explanation
Description Check Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check Passed The title 'refactor: reduce wait-for-action to one PR monitor' matches the main change. The PR replaces the multi-mode wait-for-action dispatcher with a single read-only Python PR monitor, deletes the…
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.
Full details: Title check

Explanation

The title 'refactor: reduce wait-for-action to one PR monitor' matches the main change. The PR replaces the multi-mode wait-for-action dispatcher with a single read-only Python PR monitor, deletes the CodeRabbit and PR-Agent watchers and their tests, and updates the README and skill docs to the single entrypoint. The title is specific and concise and follows conventional commit style.

Finishing Touches

Generate unit tests (beta)

  • Commit to this branch
  • Create a new PR
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

Recent review info

Run configuration

Commits

Reviewing files that changed from the base of the PR and between 192f87c and ba2352e.

Files selected for processing: 19

  • .claude-plugin/marketplace.json
  • .cursor-plugin/marketplace.json
  • .github/plugin/marketplace.json
  • README.md
  • skills/independent-review/.cursor-plugin/plugin.json
  • skills/independent-review/SKILL.md
  • skills/independent-review/agents/openai.yaml
  • skills/lets-finish/SKILL.md
  • skills/wait-for-action/.cursor-plugin/plugin.json
  • skills/wait-for-action/SKILL.md
  • skills/wait-for-action/scripts/coderabbit.py
  • skills/wait-for-action/scripts/pr_agent.py
  • skills/wait-for-action/scripts/wait-for-action.sh
  • skills/wait-for-action/scripts/wait_for_action.py
  • tests/test_coderabbit.py
  • tests/test_pr_activity.py
  • tests/test_pr_agent.py
  • tests/test_wait_for_checks.py
  • tests/test_wait_for_ready.py
Files with no reviewable changes: 7

  • skills/wait-for-action/scripts/coderabbit.py
  • skills/wait-for-action/scripts/pr_agent.py
  • skills/wait-for-action/scripts/wait-for-action.sh
  • tests/test_coderabbit.py
  • tests/test_pr_agent.py
  • tests/test_wait_for_checks.py
  • tests/test_wait_for_ready.py

Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.

@archae0pteryx

Copy link
Copy Markdown
Contributor Author

Independent review completed in two fresh-context Code Reviewer passes. No blockers were reported. Caller guidance and descriptions were corrected, reviewer identity verification was made explicit, the default artifact directory now uses the system private temporary-directory facility, and remote text is flattened before printing to prevent status-line/terminal-control injection. These follow-ups have regression proof; no fresh no-findings verdict is claimed for the final follow-up.

API volume is documented; additional caching, provider-specific retry paths, wrappers, and tuning modes were declined to preserve the requested single opinionated function. Standard argparse usage exit codes and the documented Python 3.10+ runtime are retained. No release/version bump is included.

112 tests pass; generated manifests validate.

@archae0pteryx
archae0pteryx merged commit 61f1067 into main Nov 15, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant