How do I install the Unstick Queue skill?

Run npx skillmds@latest add omninode-ai/unstick-queue in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Unstick Queue safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Unstick Queue?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Unstick Queue free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Unstick Queue?

OmniNode-ai (@omninode-ai) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Unstick Queue by omninode-ai — Agent Skill

Name: Unstick Queue
Author: OmniNode-ai

File contents

unstick_queue — auto-unwedge merge-queue stalls

Purpose

GitHub merge queues require ALL statusCheckRollup entries to complete — not just the repo's required checks on main. Third-party bot check-runs, flaky runners and dropped webhooks can hang IN_PROGRESS indefinitely, pinning the queue head AWAITING_CHECKS forever and blocking every downstream PR from merging.

This skill detects that specific condition and recovers by dequeuing + re-enqueuing the head PR. It is deliberately scoped to avoid admin-bypass of genuine failures: a PR with a failed required check is classified BROKEN and left alone.

Scope

The skill's scope is defined as follows:

Probe repository.mergeQueue.entries for every repo in scope.
For each entry at position == 1 with state == AWAITING_CHECKS longer than awaiting-minutes (default 30):
- Fetch statusCheckRollup via GraphQL.
- Run plugins/onex/hooks/lib/queue_stall_classifier.classify_queue_entry.
Act on the verdict:
- STALL — dequeue via dequeuePullRequest, sleep briefly, re-enqueue via enqueuePullRequest. Record the event under $ONEX_STATE_DIR/queue-unstick/<repo>/pr-<N>.json.
- BROKEN — skip (real failure; not our problem).
- ESCALATE — same PR was unstuck ≥ 3 times within the past hour; stop auto-healing and emit friction via /onex:record_friction so the deeper cause is surfaced.
- HEALTHY — no action.

Non-goals

Do NOT admin-bypass check-runs (mark complete, skip).
Do NOT edit branch-protection required contexts.
Do NOT close/reopen PRs.

Invocation

Tick path (preferred): ai.omninode.unstick-queue.plist runs every 10 minutes via scripts/cron-unstick-queue.sh.

Interactive: /onex:unstick_queue --dry-run --repos omnibase_infra,omniclaude for manual inspection.

Files

plugins/onex/hooks/lib/queue_stall_classifier.py — pure classifier + persistence (tested).
scripts/lib/run-unstick-queue.py — the runner the cron wrapper invokes.
scripts/cron-unstick-queue.sh — headless wrapper + lock + circuit breaker.
scripts/launchd/ai.omninode.unstick-queue.plist — launchd agent (600s interval).

---name: unstick-queuedescription: Detect merge-queue head PRs stalled by orphaned third-party check-runs and auto-dequeue+re-enqueue to unwedge downstream PRs---# unstick_queue — auto-unwedge merge-queue stalls## PurposeGitHub merge queues require ALL `statusCheckRollup` entries to complete — not just the repo's required checks on `main`. Third-party bot check-runs, flaky runners and dropped webhooks can hang `IN_PROGRESS` indefinitely, pinning the queue head `AWAITING_CHECKS` forever and blocking every downstream PR from merging.This skill detects that specific condition and recovers by dequeuing + re-enqueuing the head PR. It is deliberately scoped to avoid admin-bypass of genuine failures: a PR with a failed *required* check is classified `BROKEN` and left alone.## ScopeThe skill's scope is defined as follows:1. Probe `repository.mergeQueue.entries` for every repo in scope.2. For each entry at `position == 1` with `state == AWAITING_CHECKS` longer than `awaiting-minutes` (default 30): - Fetch `statusCheckRollup` via GraphQL. - Run `plugins/onex/hooks/lib/queue_stall_classifier.classify_queue_entry`.3. Act on the verdict: - `STALL` — dequeue via `dequeuePullRequest`, sleep briefly, re-enqueue via `enqueuePullRequest`. Record the event under `$ONEX_STATE_DIR/queue-unstick/<repo>/pr-<N>.json`. - `BROKEN` — skip (real failure; not our problem). - `ESCALATE` — same PR was unstuck ≥ 3 times within the past hour; stop auto-healing and emit friction via `/onex:record_friction` so the deeper cause is surfaced. - `HEALTHY` — no action.## Non-goals- Do NOT admin-bypass check-runs (mark complete, skip).- Do NOT edit branch-protection required contexts.- Do NOT close/reopen PRs.## InvocationTick path (preferred): `ai.omninode.unstick-queue.plist` runs every 10 minutes via `scripts/cron-unstick-queue.sh`.Interactive: `/onex:unstick_queue --dry-run --repos omnibase_infra,omniclaude` for manual inspection.## Files- `plugins/onex/hooks/lib/queue_stall_classifier.py` — pure classifier + persistence (tested).- `scripts/lib/run-unstick-queue.py` — the runner the cron wrapper invokes.- `scripts/cron-unstick-queue.sh` — headless wrapper + lock + circuit breaker.- `scripts/launchd/ai.omninode.unstick-queue.plist` — launchd agent (600s interval).## See Also- Sibling: `_queue_heal` in `cron-merge-sweep.sh` (handles method-mismatch silent drops — this skill handles check-rollup stalls)

Frequently asked questions

Unstick Queue