MCP.Market

gh-issues skill

by openclaw·openclaw/openclaw·394k stars·MIT

Fetch GitHub issues, select candidates, spawn background fix agents, open PRs, and optionally process PR review comments.

A100/100content scan

Is the gh-issues skill safe?

Clean: nothing in its files matched our rules. We read 1 file in the folder on 2026-10-25.

No findings.

Install the gh-issues skill

A skill is a folder. Copy it into your agent's skills folder and the agent loads it when the task matches its description.

git clone --depth 1 https://github.com/openclaw/openclaw.git /tmp/openclaw
mkdir -p ~/.claude/skills
cp -r /tmp/openclaw/skills/gh-issues ~/.claude/skills/gh-issues
available in every project

In the Claude apps, zip the folder and upload it from the Skills settings. The folder on GitHub

The instructions your agent would load

SKILL.md as published, without the frontmatter. Read it on GitHub

gh-issues

Use for issue-to-PR automation. Prefer gh CLI; fall back to gh api only when a high-level command lacks the needed field.

Arguments

  • positional owner/repo: optional; else infer from git remote get-url origin.
  • --label : filter.
  • --limit : default 10.
  • --milestone : filter.
  • --assignee : filter.
  • --state open|closed|all: default open.
  • --fork : push branches to fork, PR to source.
  • --watch: poll issues + reviews.
  • --interval : default 5.
  • --dry-run: list only.
  • --yes: no confirmation.
  • --reviews-only: skip issue fixing; handle PR reviews.

Phase 1: resolve repo

git remote get-url origin
if [ -z "${GH_TOKEN:-}" ]; then
  CONFIG_PATH="${OPENCLAW_CONFIG_PATH:-${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/openclaw.json}"
  GH_TOKEN=$(jq -r '.skills.entries["gh-issues"].apiKey // empty' "$CONFIG_PATH" 2>/dev/null || true)
  if [ -n "$GH_TOKEN" ]; then export GH_TOKEN; fi
fi
gh auth status
gh repo view OWNER/REPO --json nameWithOwner,defaultBranchRef

If gh auth status fails and GH_TOKEN is missing, stop and ask for GitHub auth/config.

Derived:

  • SOURCE_REPO: issue repo.
  • PUSH_REPO: fork if set, else source.
  • BASE_BRANCH: source default branch unless user says otherwise.
  • PUSH_REMOTE: fork in fork mode, else origin.

Stop on dirty worktree unless user confirms that workers should ignore uncommitted changes.

In fork mode, do not mutate remotes before confirmation or during --dry-run.

Verify auth/read access only:

gh auth token >/dev/null || test -n "${GH_TOKEN:-}"
gh repo view "$PUSH_REPO" --json nameWithOwner
git ls-remote --exit-code origin HEAD

Phase 2: fetch issues

Build filters and fetch:

gh issue list --repo "$SOURCE_REPO" --state open --limit 10 --json number,title,labels,url,body,assignees,milestone

Add --label, --milestone, --assignee, --state, --limit as requested. gh issue list already excludes PRs.

If none found: report no matches. If --dry-run: show compact list and stop.

Phase 3: avoid duplicate work

For each candidate:

gh pr list --repo "$SOURCE_REPO" --search "$SOURCE_REPO#<n>" --state open --json number,url,title,headRefName
gh pr list --repo "$SOURCE_REPO" --head "fix/issue-<n>" --state open --json number,url
gh api "repos/$PUSH_REPO/branches/fix/issue-<n>" >/dev/null

Skip candidates with an open PR, existing branch, or active local claim.

Claim file:

${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/gh-issues-<owner>-<repo>.json

Expire claims older than 2 hours. Create the parent directory before writing.

Phase 4: confirm

Unless --yes or --cron, ask user to choose:

  • all
  • comma-separated issue numbers
  • cancel

After confirmation, in fork mode, configure the push remote before handing work to agents:

gh auth setup-git
git remote get-url fork || git remote add fork "https://github.com/$PUSH_REPO.git"
git remote set-url fork "https://github.com/$PUSH_REPO.git"
git ls-remote --exit-code fork HEAD

Phase 5: spawn workers

Launch up to 8 background workers. Do not block on each worker when --cron.

Before each spawn, write a claim for SOURCE_REPO# with the current ISO timestamp. After a worker reports PR/failure, remove or update the claim. This prevents watch/cron overlap before a branch or PR exists.

Worker prompt must include:

  • issue URL, title, body, labels.
  • SOURCEREPO, PUSHREPO, BASEBRANCH, PUSHREMOTE, fork mode.
  • target branch fix/issue-.
  • required proof and PR body.
  • notification route.

Worker instructions:

Use gh and git. Do not handwave.
Checkout/create fix/issue-<n> from BASE_BRANCH.
Implement minimal fix.
Run relevant tests.
Commit with conventional message.
Push to PUSH_REMOTE.
Open PR against SOURCE_REPO BASE_BRANCH.
PR body: What Problem This Solves + Why This Change Was Made + User Impact + Evidence + visible Fixes SOURCE_REPO#<n>.
Report PR URL or failure reason.
Send completion/failure with openclaw message send if route provided.

Use coding-agent launch rules when available.

Phase 6: collect

Poll workers with process or task registry. Report:

  • issue number + title.
  • status: PR opened, skipped, failed, timed out.
  • PR URL or reason.

Notify channel only with final compact summary.

Reviews-only / watch reviews

Discover open PRs:

gh pr list --repo "$SOURCE_REPO" --state open --json number,title,url,headRefName,reviewDecision \
  --jq '[.[] | select(.headRefName | startswith("fix/issue-"))]'

Fetch review threads/comments:

gh pr view <n> --repo "$SOURCE_REPO" --json url,headRefName,comments,reviews
gh api "repos/$SOURCE_REPO/pulls/<n>/comments"
gh api "repos/$SOURCE_REPO/issues/<n>/comments"

Only process fix/issue-* PRs created by this workflow unless the user explicitly named PR numbers. Group actionable comments by PR. Ignore praise, status, duplicates, and already-addressed comments. Spawn one worker per selected/scoped PR, same background rules.

Review worker instructions:

Checkout PR branch.
Read all actionable review comments.
Patch minimal changes.
Run relevant tests.
Commit and push normally; do not force-push unless explicitly told.
Reply to addressed comments with fix + commit/file reference.
Report comments addressed/skipped and proof.

Watch mode

Loop:

  1. Fetch issues.
  2. Spawn eligible issue workers.
  3. Process actionable PR reviews.
  4. Sleep --interval.
  5. Stop when user says stop.

Keep cumulative summary small.

More skills from openclaw/openclaw

  • A1passwordSet up and use 1Password CLI for sign-in, desktop integration, and reading or injecting secrets.
  • Aacp-routerRoute plain-language requests for Claude Code, Cursor, Copilot, OpenClaw ACP, OpenCode, Gemini CLI, Grok Build, Qwen, Kiro, Kimi, iFlow, Factory Droid, Kilocode, or explicit ACP harness work into either OpenClaw ACP runtime sessions or direct acpx-driven sessions ("telephone game" flow). For coding-agent thread requests, read this skill first, then use only `sessions_spawn` for thread creation. Codex chat binding defaults to the native Codex app-server plugin unless ACP is explicit or background spawn needs ACP.
  • Aadd-model-providerAdd and live-prove a model provider with non-interactive config one-liners, without exposing credentials.
  • Aagent-transcriptRequested GitHub PR/issue agent transcripts: redact, trim, preview, and insert safely.
  • Aapple-notesCreate, view, edit, delete, search, move, or export Apple Notes via the memo CLI on macOS.
  • Aapple-remindersList, add, edit, complete, or delete Apple Reminders and reminder lists via remindctl.
  • Aauto-qaRun OpenClaw-wide autonomous QA and live/stress campaigns across independent subsystem lanes, with verified fixes and a resumable evidence report.
  • AautoreviewStructured code review through the shared OpenClaw agent-skills installation.
  • Abear-notesCreate, search, and manage Bear notes via grizzly CLI.
  • Ablock-kitUse proactively for structured or interactive Slack replies, and when asked to author or validate native Slack Block Kit JSON.
  • AblogwatcherMonitor blogs and RSS/Atom feeds for updates using the blogwatcher CLI.
  • AblucliBluOS CLI (blu) for discovery, playback, grouping, and volume.

All agent skills → · MCP servers