Guide · Agent skill
How to use the ClawSweeper Status Check agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The clawsweeper-status skill gives an AI agent a ready-made status report for the ClawSweeper review bot that runs on the openclaw GitHub repositories. It is aimed at maintainers and engineers on the openclaw team who need to see whether automated reviews are stalled, overloaded or deliberately holding work. This guide explains what the skill does, how to install it, how to ask for a report, how to read the output and what to be careful about before running it.
ClawSweeper Status Checksteipete/agent-scripts · MIT Download the skill Quick start
- 1Download and unzip the skill. Download the ZIP of clawsweeper-status from the page where you found it and unzip it. You should see a clawsweeper-status folder containing SKILL.md, agents/openai.yaml and the scripts folder.
- 2Place the folder where the agent can find it. Copy the clawsweeper-status folder into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Set up the tools and access. Make sure bash, the GitHub CLI (gh) and jq are installed, and that gh is logged in with access to openclaw/clawsweeper and openclaw/openclaw.
- 4Edit the hard-coded paths. Open scripts/clawsweeper-status.sh and replace the author's local paths with the correct location on your machine.
- 5Ask for a status report. Request a ClawSweeper status in plain language, or run the script with --hours, --repo or --limit. Review the queue health section first, since it leads whenever the state is not healthy.
What the skill makes the agent do
When you ask the agent for a ClawSweeper status, it runs a bundled shell script (scripts/clawsweeper-status.sh). The script uses the GitHub CLI (gh) to collect recent merged, reviewed, commented and closed items, active workflow runs, and capacity and queue health.
The agent then reports the results in fixed sections. It is instructed to explain whether a backlog is deliberate holding, rate limiting, or parked items that need a human, rather than just listing numbers.
- Workers: workflow state, Codex jobs against configured capacity, and active workflow groups
- Queue health: shown first whenever it is not healthy
- Publication tail: pending, ready, backoff, parked and active counts, plus retry and parked reasons
- Recently merged, reviewed, commented and closed: full GitHub URLs with one-line summaries
Install the clawsweeper-status skill
Download the ZIP of the skill from the page where you found it, then unzip it. Place the resulting clawsweeper-status folder in one of two locations: your personal skills folder at ~/.claude/skills/, or a project's .claude/skills/ folder if you want the skill available only inside that project.
If you use the Claude apps instead, you can upload the skill under Settings → Capabilities → Skills. Either way, the skill folder should contain SKILL.md, the agents/openai.yaml file, and the scripts folder.
- Personal install: ~/.claude/skills/clawsweeper-status/
- Project install: .claude/skills/clawsweeper-status/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload the skill
Check the requirements before you run it
The script needs several tools and access to specific repositories. Without them the report will fail or be incomplete. The skill's queries are read-only, but they do need an authenticated GitHub session.
Note that the script is hard-wired to openclaw repositories. It only reports on ClawSweeper and Codex-related workflow activity, so it is not a general GitHub status checker.
- bash
- GitHub CLI (gh), authenticated with access to the target repositories
- jq
- GitHub API access to openclaw/clawsweeper and openclaw/openclaw (read queries only)
Ask for a status report with natural requests
You do not need to type the script path yourself. Ask the agent in plain language, for example for a ClawSweeper status, or whether automated reviews look stalled. The skill tells the agent to run the status script when status is requested.
If you run the script directly, the source shows these options. The paths in the source point to the author's own machine, so replace them with the location where the skill is installed on your system.
- --hours sets the time window (the source example uses 10 hours for the default target, openclaw/openclaw)
- --repo selects a different target repository (the source example uses openclaw/clawhub)
- --limit sets how many rows each activity section shows (the source example uses 15)
/Users/steipete/Projects/agent-scripts/skills/clawsweeper-status/scripts/clawsweeper-status.sh --hours 10Read the queue health section correctly
Pending depth on its own is not a verdict. Read the split the script prints underneath the queue health line, because it tells you whether work is being held back, rate limited or parked.
Parked items do not recover on their own. Per the skill's guidance, they retry at 5, 10 and 20 minutes and then wait for a human, so they need to be called out explicitly.
- ready or admissible near zero while pending is deep: the lane is deliberately holding items back, not stalled
- Queue backoff: throttle_retry N: GitHub is rate limiting, and the lane recovers once quota returns
- Queue parked (needs operator): review_retry_exhausted N: does not self-heal and needs a person
- Shed since reset climbing into the thousands: sustained overload, not a blip
- occupancy unavailable does not mean zero
Safety notes before running the script
This skill runs scripts and uses the network. It calls the GitHub API through gh, and the source says it only performs read queries against openclaw/clawsweeper and openclaw/openclaw. Read the bundled scripts yourself before running them, and make sure you are comfortable with the GitHub credentials your gh session uses.
The script has hard-coded local paths from the author's machine and queue data specific to that setup. Expect to edit paths before the skill works on your machine, and treat the output as specific to the openclaw setup.
- Review scripts/clawsweeper-status.sh before first use
- Check which GitHub account gh is logged into
- Edit hard-coded paths before running on a different machine
- The bundled test script is scripts/clawsweeper-status.test.sh, with sample data in scripts/fixtures/
Tips for getting useful reports
Treat cancelled repository-dispatch review runs as usually expected, since they are often superseded by a newer event for the same item. Count Codex usage from actual in-progress or queued jobs, and treat pending workflow runs as concurrency waiters rather than queued Codex jobs.
If worker counts disagree, compare the status-filtered gh run list results with the default recent-run list. The skill also notes that large workflow snapshots still reach all activity sections, and that required fetch or parser failures are fatal, while optional queue telemetry may show as unavailable.
- Increase --limit when activity sections look truncated
- Expect full GitHub URLs in the final answer
- Missing health counts are unknown, not zero
- Active workflow groups are capped at 20, ordered by descending run count
What people use it for
- Checking each morning whether the ClawSweeper review bot is keeping up with new openclaw pull requests
- Telling a deliberate queue hold apart from rate limiting or parked items that need a human
- Reviewing which openclaw PRs were recently merged, reviewed, commented on or closed, with full URLs
- Investigating a suspected overload by watching the shed-since-reset counter over time
Frequently asked questions
Is the clawsweeper-status skill free to use?
The skill is published under the MIT licence, as stated on its listing. You still need your own GitHub access through the gh CLI, and the skill's queries are read-only against openclaw/clawsweeper and openclaw/openclaw.
What do I need installed to run clawsweeper-status?
You need bash, the GitHub CLI (gh) authenticated with access to the target repositories, and jq. The skill also needs GitHub API access to openclaw/clawsweeper and openclaw/openclaw for read queries.
Where do I install the skill in Claude?
Put the clawsweeper-status folder in ~/.claude/skills/ for personal use, or in a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Why does the script fail or show wrong paths on my computer?
The script contains hard-coded local paths from the author's own machine. Replace those paths with the location of the skill on your system before running it, and confirm that gh is authenticated to an account with access to the openclaw repositories.
Can I use clawsweeper-status for any GitHub repository?
No. The script is hard-wired to openclaw repositories and only reports on ClawSweeper and Codex-related workflow activity, so it is not a generic GitHub status check. The source does not name alternatives; for general GitHub checks you can use gh directly, which the skill itself relies on.
What does 'Queue parked (needs operator)' mean?
It means items have exhausted their automatic retries (review_retry_exhausted). Per the skill's guidance, they retry at 5, 10 and 20 minutes and then wait for a human, so they will not recover on their own.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use clawsweeper-status if you maintain or work on the openclaw projects and need a quick, structured view of whether ClawSweeper reviews are stalled, rate limited or overloaded. Pick something else if you work outside the openclaw GitHub organisation or need a general repository status check, since the script is tied to openclaw repositories and the author's own machine setup.
Download the skill