watch

Monitors GitHub CI runs across branches and acts on failures until all checks pass.

1|Updated May 6, 2026
One-click install
npx skills add https://github.com/diy-accounting-uk/submit.diyaccounting.co.uk --skill watch-diy-accounting-uk
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: watch
Source: https://github.com/diy-accounting-uk/submit.diyaccounting.co.uk/tree/main/.claude/skills/watch
Command: npx skills add https://github.com/diy-accounting-uk/submit.diyaccounting.co.uk --skill watch-diy-accounting-uk

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Keeping a repository's CI green requires constant polling of GitHub Actions runs across main and every open pull request, which blocks the session and wastes attention. This Skill arms a persistent background monitor that reports failures and merge-ready PRs as events, so work continues while CI runs. ## Core Features & Use Cases - Background CI monitoring: Arms a persistent Monitor over scripts/watch-ci.sh that polls every 75 seconds and emits one line per newly failed run, mergeable PR, or final tally. - Failure triage guidance: Provides commands to open failed runs, fetch job logs via the GitHub API, and diagnose whole-run failures together before fixing. - Merge readiness probing: Detects PRs whose latest push and pull_request workflow runs are all terminal and none failed, then routes them to the /auto-merge skill. - Use Case: After pushing a branch and opening a PR, invoke the watch to keep tracking CI across main and all open PRs, get notified of a red deploy run, diagnose the failing job log, fix the right layer, and push once until every check is green. ## Quick Start Watch the builds on this repository and keep every branch green until all CI checks pass.

Frequently Asked Questions about watch

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

FAQPage Schema
How do I monitor GitHub Actions CI runs in the background?▼

Arm a persistent Monitor tool over a polling script that runs gh run list every 75 seconds and prints one line per newly failed run or merge-ready PR. Each stdout line becomes a notification, so the session stays free while CI runs.

How to diagnose a failed GitHub Actions run with the gh CLI?▼

Open the run with gh run view <run-id> --json jobs to list failed steps, then fetch the job log via gh api repos/<owner>/<repo>/actions/jobs/<job-id>/logs, strip ANSI codes, and grep for errors. Gather the whole run's failures before fixing anything.

Which GitHub workflow run conclusions count as failures?▼

failure, timed_out, action_required, and startup_failure are treated as red. success, skipped, cancelled, and neutral do not block a PR, since skips can come from paths filters and cancellations are usually supersessions.

Why does a cancelled GitHub Actions run not mean a defect?▼

A cancelled run is usually a supersession by a newer push, so check for a later run in the same concurrency group. With cancel-in-progress false, GitHub can also drop a queued run when a third arrives, silently skipping a deploy.

When is it safe to push to a branch with deploy workflows?▼

Never push while any of that branch's deploy runs are in flight, confirmed by reading the runs rather than assuming elapsed time. A cancelled deploy mid-change can leave infrastructure partially applied.