Gh Issues

GitHub Issues Watcher

Purpose

Process every open GitHub issue that is unassigned or assigned to the current user (@me), one after another, by delegating each to the gh-issue skill. Stop on first hard failure so it can be inspected.

Args

• --limit N — process at most N issues this run (default: all).
• --label foo — only issues carrying label foo.
• --dry-run — list issues that would be processed; do not invoke gh-issue.

Strip leading # if user passes #123 style.

Phase 1: Discover

Fetch open issues that are unassigned or assigned to @me, oldest first. GitHub search does not OR these cleanly, so run two queries and merge:

# Unassigned
gh issue list \
  --state open \
  --search "no:assignee" \
  --json number,title,labels,assignees,createdAt \
  --limit 200

# Assigned to me
gh issue list \
  --state open \
  --assignee "@me" \
  --json number,title,labels,assignees,createdAt \
  --limit 200

Merge:

• Deduplicate by number.
• Keep only issues whose assignees array is empty or contains the current user (gh api user -q .login).
• Drop issues assigned to anyone else (defensive).
• Apply --label filter if given.
• Apply --limit if given.
• Sort ascending by createdAt (FIFO).

Print the queue: # [assignee] per line, where [assignee] is unassigned or @me. If empty, exit cleanly.

Phase 2: Worktree Sanity

Before touching any issue:

git status --porcelain
git fetch origin main
git checkout main && git reset --hard origin/main

Abort if worktree dirty. Never auto-stash.

Phase 3: Process Loop

For each issue in the queue:

1. Re-check assignment state (someone else may have grabbed it):

``bash gh issue view --json assignees -q '.assignees[].login' me=$(gh api user -q .login) ``

• Empty output → unassigned, proceed.
• Only $me listed → already mine, proceed (skip self-assign step).
• Any other login present → skip this issue.

1. Invoke the gh-issue skill with the issue number. That skill owns:

• self-assign via gh issue edit --add-assignee @me (no-op if already assigned).
• branch from fresh main (prefix from labels: fix/, feat/, docs/).
• TDD implementation.
• go test ./... green locally.
• regen of derived artifacts (schema / sync / playground) when applicable.
• changelog entry under ## [Unreleased].
• commit with Related to #.
• mandatory final refactor pass over every touched file (separate ref(...) commit) before opening the PR.
• PR opened with --assignee Chemaclass, matching label, Closes # in body.

1. After gh-issue returns, wait for CI green on the PR:

``bash gh pr checks --watch `` Fix red checks on the branch before moving on.

1. Merge when allowed:

``bash gh pr merge --squash --admin --delete-branch ` Fall back to --auto --squash --delete-branch if --admin` is rejected.

1. Sync main for next iteration:

``bash git checkout main && git fetch origin main && git reset --hard origin/main ``

1. Continue with next issue.

Stop Conditions

Halt the loop and surface the failure when:

• gh-issue errors out or leaves the worktree dirty.
• go test ./... fails after implementation.
• agnostic-ai sync --check drifts after spec/adapter edits.
• CI stays red after one fix attempt.
• Merge is blocked by branch protection beyond --admin bypass.
• --limit reached.
• Queue empty.

Do not retry blindly. Report which issue failed and why.

Dry Run

With --dry-run, only execute Phase 1 and print the queue. No assignment, no branching, no commits.

Preconditions

• gh authenticated, can read issues, open and merge PRs.
• Worktree clean.
• main exists and tracks origin/main.
• gh-issue skill available in this session.

Notes

• Treat GitHub CI as the full quality gate; locally run focused tests during implementation, full go test ./... once before commit.
• Never split bundled changes into multiple PRs unless the issue explicitly demands it.
• One concern per PR (.claude/rules/go-style.md). Open separate PRs for refactors that the issue did not ask for.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Chemaclass
• Source: Chemaclass/agnostic-ai
• License: MIT
• Homepage: https://chemaclass.github.io/agnostic-ai/

Install and usage instructions live in the source repository linked above.