ot-pr-autopilot

Diagnose an open PR's state and chain the matching review, fix, QA, and merge skills.

Updated Sep 10, 2026
One-click install
npx skills add https://github.com/triage-software/skills --skill ot-pr-autopilot-triage-software
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ot-pr-autopilot
Source: https://github.com/triage-software/skills/tree/main/skills/ot-pr-autopilot
Command: npx skills add https://github.com/triage-software/skills --skill ot-pr-autopilot-triage-software

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Open pull requests stall in unclear states — unfinished plan steps, missing reviews, red CI, unresolved conversations, or absent QA evidence — and figuring out what remains requires manual inspection. This Skill diagnoses one PR's actual state and runs the right sequence of companion skills to drive it toward merge-ready. ## Core Features & Use Cases - Ten-signal diagnosis: Reads identity, plan progress, diff scope, review decision, unresolved conversations, CI status, mergeability, labels, QA evidence, and claim state into a single PR State Report. - State-matrix chaining: Maps the diagnosis to an ordered chain of companion skills (continue implementation, autofix, review, UI QA, follow-up issues, merge) and executes them verbatim, re-checking cheap signals between steps. - Safe autonomy controls: Supports --dry-run, --confirm, --allow-merge (off by default), --force, and --max-iterations, with a three-signal claim lock, stale-lock recovery, and a bounded CI wait that reports before waiting. - Use Case: Run it on PR 4321 after returning from the weekend; it detects pending plan steps and red CI, finishes the implementation, fixes the failing checks, captures QA screenshots, and posts a consolidated status report. ## Quick Start Ask the agent to drive PR 4321 to completion with ot-pr-autopilot, optionally adding --dry-run first to see the proposed chain without mutating anything.

Frequently Asked Questions about ot-pr-autopilot

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I finish an open pull request automatically?▼

Run ot-pr-autopilot with the PR number, for example "finish PR 123". It diagnoses the PR's state across ten signals, then runs the matching chain of companion skills — continue, autofix, review, QA — and publishes a status report.

What does the PR autopilot check before acting on a pull request?▼

It collects ten read-only signals: identity, plan progress, diff scope, review decision, unresolved conversations, CI status, mergeability, labels, QA evidence, and claim state. These feed a state matrix that decides which chain steps to run.

Can ot-pr-autopilot merge a pull request on its own?▼

No, merging is off by default. The chain stops at merge-ready unless --allow-merge is passed and the QA gate is satisfied; the actual merge is delegated to the ot-approve-merge-pr skill.

Does it work on pull requests from forks or other authors?▼

Yes, but another author's PR gets review plus handoff rather than autofix, unless the user explicitly requests the autofix chain. Your own fork PR is pushable and is driven like a same-repo PR.

What happens if CI is still running when the run finishes?▼

The summary comment, labels, and lock release are posted before any CI wait, so nothing is stranded. The wait is capped at ci.maxWaitMinutes (default 40); on exhaustion it reports local validation results and names the pending checks.

When should I use ot-auto-fix-pr instead of ot-pr-autopilot?▼

Use ot-auto-fix-pr directly when you already know the PR needs the merge-readiness loop. Use ot-pr-autopilot when you do not know the PR's state and need diagnosis to decide which skills to run.