Guide · Agent skill
How to use the ClawHub ban appeals agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
clawhub-ban-appeals is an agent skill, meaning a folder with a SKILL.md instruction file that an AI agent such as Claude reads before doing a task. This one guides the agent through reviewing an appeal from a ClawHub user whose account was banned or held, so an authorized moderator can reach a consistent, evidence-backed decision. This guide shows you how to install it, how to ask for a review, what it needs to run, and where its limits are.
ClawHub ban appealsopenclaw/clawhub · MIT Download the skill Quick start
- 1Install the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings, Capabilities, Skills in the Claude apps.
- 2Confirm the companions are available. Make sure the clawhub-moderation and clawhub-clawscan skills and their referenced documents are also installed and that you have ClawHub staff access.
- 3Give the appeal reference and account. Provide the exact account handle or id and the appeal or source reference so the agent can read the full thread.
- 4Ask for the account state and evidence. Ask the agent to confirm the current ban or hold, its recorded reason and time, and the affected listings or versions, with sources and capture times.
- 5Review the recommendation. Read the single recommendation the agent returns, along with its uncertainties and any decision that must stay with an authorized moderator.
- 6Approve any action explicitly. Only approve a specific admin command with its exact target and reason. Otherwise the agent prepares it for Admin's approval path.
What the skill makes the agent do
The skill tells the agent to confirm the exact account state before judging anything. It separates an account ban, a moderation hold, a hidden listing or blocked release, and a namespace or content-rights dispute, because each one has a different remedy. It then checks the evidence behind the action and recommends one of four outcomes: uphold with reasons, restore the specified state under moderation authority, request the exact missing evidence, or escalate for security or legal review.
The skill also keeps scanner problems separate from the account decision. If a scanner finding is validated as inaccurate, the agent routes it to the ClawScan upstream reporting process rather than treating the report as a restoration.
Install the clawhub-ban-appeals skill
Download the skill ZIP from this page and unzip it. For a personal setup, place the unzipped folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder. Alternatively, in the Claude apps, open Settings, then Capabilities, then Skills, and upload the skill there.
After installing, start a new session so the agent can see the skill. The skill is licensed under MIT.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload
Check the requirements before you start
This skill is not standalone. It needs access to ClawHub authenticated staff read tools and admin commands, which the agent reaches through the clawhub-moderation skill. It also depends on the companion skills clawhub-moderation and clawhub-clawscan, and on the reference documents they point to. The skill's own files are a single SKILL.md. It runs no scripts, but it does use the network.
If you do not have staff access, the agent can still prepare a review, but it cannot verify the live account state. The skill treats missing access as a verification limitation, not as proof of anything.
- ClawHub staff read tools and admin commands (through clawhub-moderation)
- Companion skills: clawhub-moderation and clawhub-clawscan, with their referenced documents
- Scripts: none. Network: yes
Ask for a review with a natural request
You do not need special syntax. Give the agent the account handle or id, the appeal reference, and the question you want answered. The agent is expected to start from the exact handle or id and the existing appeal or source reference, and to read the full appeal thread and prior decisions.
A good request names the appeal, the account, and the outcome you need. Make clear whether you want a recommendation only, or whether you are authorizing a specific action.
- Review appeal APP-123 for account handle example-user and tell me whether the ban is still active.
- Read the full thread for this appeal and check whether it duplicates a resolved case.
- Separate any moderation hold on example-user from the account ban and recommend an outcome.
- The ClawScan finding on my package looks wrong. Check whether it is a validated inaccuracy and prepare the upstream report draft.
Understand how the review works
The agent records the source and capture time of each observation. A public missing page or a failed login is not sufficient evidence of a ban. It matches each relied-on scan to the exact submitted artifact and version, and it inspects every recorded ground for the action. Refuting one finding does not automatically entitle the account to restoration, and a newer corrected version does not show the original decision was wrong.
Before any action, the agent identifies which remedy applies to which state. Remedies differ: unban restores eligible content from the matching ban flow but leaves revoked API tokens revoked, while lifting a moderation hold resolves only that hold.
- Account ban: review the recorded action and its evidence. users unban is the restoration path when authorized.
- Account moderation hold: review separately. users lift-moderation-hold is distinct from unban.
- Hidden listing or blocked release: use listing and report moderation guidance.
- Namespace or content-rights dispute: route to the existing claims process, not the ban appeal flow.
Keep actions tied to explicit authorization
The skill grants no standing authority to restore accounts, send correspondence, or publish scanner reports. Every admin command needs the exact target, the reason, and existing explicit authorization. If that authorization is absent, the agent prepares the command for Admin's approval path instead of running it.
Reclassifying a ban changes its stored reason but does not unban the account. skills unhide applies a manual override for a justified listing decision and is not a scanner repair or a way to clear an account action. Outbound email follows moderation's signoff rules, and the agent prepares a draft when it lacks send authority.
What people use it for
- Checking whether a banned account's ban, hold, and blocked releases are separate issues before any decision is made.
- Preparing a consistent written recommendation for a moderator who must uphold, restore, request evidence, or escalate.
- Deduplicating an appeal against prior decisions and confirming whether it is already resolved.
- Validating a suspected ClawScan inaccuracy and preparing an upstream report in openclaw/clawscan that stays separate from the account decision.
Frequently asked questions
Is the clawhub-ban-appeals skill free to use?
The skill is licensed under MIT. Using it still requires ClawHub staff access through the clawhub-moderation skill, so it is not useful on its own for people without that access.
How do I install the clawhub-ban-appeals skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder for one project. In the Claude apps, you can instead upload it under Settings, Capabilities, Skills.
What does the clawhub-ban-appeals skill require to run?
It requires access to ClawHub authenticated staff read tools and admin commands through the clawhub-moderation skill. It also needs the companion skills clawhub-moderation and clawhub-clawscan and their referenced documents. It runs no scripts but uses the network.
Can this skill restore a banned account automatically?
No. The skill says restoration and final calls must stay with an authorized moderator. It has no standing authority to restore an account, and it does not send emails on its own.
Why does the agent say no ban was found?
The skill requires authenticated staff reads to confirm the current account state. A public missing page or failed login is not sufficient evidence of a ban, so check that you have staff access and the correct handle or id.
Is this skill for ordinary users who want their own ban lifted?
No. It is written for authorized staff working through Admin. Users with a ban should use the published account recovery intake at https://appeals.openclaw.ai/ and not create a second intake.
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 this skill if you are an authorized ClawHub moderator or a maintainer building agent workflows around the moderation process and you want a consistent appeal checklist. Pick something else if you want an automated decision or a tool for appealing your own ban, since the skill is designed to keep final calls with authorized staff.
Download the skill