Skill Publish Vetter 🛡️#

Pre-publish privacy scan for ClawHub skills.

Every skill must pass this scan before publishing to ClawHub.

---

When to Use#

• User says "publish skill to ClawHub"
• User says "upload skill to ClawHub"
• User says "update skill on ClawHub"
• Any request to submit a skill to a public registry

---

Core Principle#

Publishing to ClawHub = publicly visible. Any leaked secret is exposed to the world.

Better safe than sorry. Even if the scan passes, always ask for user confirmation before publishing.

---

Publish Workflow (Mandatory)#

When the user requests a publish/upload/update to ClawHub, follow these steps strictly:

Step 1: Identify the Target Skill#

• If the context clearly identifies the target skill, proceed.
• Otherwise, ask the user which skill to publish.

Step 2: Run the Privacy Scan#

The scan script lives in this skill's scripts/publish-check.sh. Run it from this skill's directory:

bash

bash scripts/publish-check.sh <target-skill-directory>

It scans for:

CategoryDetection PatternExampleTokens / Keyssk-, ghp_, github_pat_, gpg_, xoxb-, glpat-, JWT tokenssk-abc123...API Keysapi_key, apikey, API_KEY, access_token with actual valuesAPI_KEY=abc123Passwordspassword, passwd, pwd with non-placeholder valuespassword=mysecret123Private KeysBEGIN.*PRIVATE KEYPEM private keyPublic KeysBEGIN.*PUBLIC KEY, ssh-rsa long stringsSSH public key**.env Values**.env files with actual (non-placeholder) assignments.env: DB_PASS=real_passwordHardcoded CredsAuthorization: Bearer with real tokenscurl -H "Authorization: Bearer sk-xxx"Personal EmailsNon-placeholder, non-org emailszhangsan@gmail.comPersonal Paths/home/username/, /Users/username/ absolute paths/Users/weidongkl/.ssh/id_rsaInternal IPs192.168.x.x, 10.x.x.x, 172.16-31.x.x``https://192.168.1.100:8080

Step 3: Output the Full Report#

Output the complete scan report to the chat. Do not summarize or omit anything.

Step 4: Secondary Confirmation (Required)#

Always ask for confirmation before publishing — even if the scan is clean.

Confirmation message must include:

1. Skill name
2. Skill directory path
3. Full scan report
4. A clear confirmation prompt

Template:

text

📋 Publish Confirmation

Skill: <name>
Path: <directory>
Version: <version>
Scan Result: <PASS / ISSUES FOUND>

[Full report here]

Reply "yes" or "confirm" to proceed with publishing, or "cancel" to abort.

Step 5: Wait for User Response#

• "yes" / "confirm" / "ok" / "go" → proceed to Step 6.
• "cancel" / "no" / "stop" → abort. Do nothing.
• No response → do not publish. Wait.

Never skip confirmation. Never auto-publish.

Step 6: Publish#

After confirmation:

bash

clawhub publish <skill-directory> --slug <slug> --name "<name>" --version "<version>" --changelog "<changelog>"

Ask the user for slug, name, version, and changelog if not provided.

Step 7: Report Result#

Tell the user whether publishing succeeded or failed.

---

Risk Levels#

LevelMeaningAction🚨 CRITICALToken, key, password, private key with actual valuesBlock publish. User must fix first.⚠️ WARNINGPersonal email, personal path, internal IPRecommend fixing before publishing.💬 INFOauthor, repository identity fields, metadata env exposureAsk user if intentionally public.

---

Red Lines (Auto-Block)#

If any of these are found, refuse to publish by default:

1. Any token with actual values (Bearer tokens, API keys, Access tokens)
2. Any private key content (PEM format, SSH private keys)
3. Any password/credential with actual values (not placeholders)
4. .env files with actual configuration values
5. Hardcoded internal IPs or domains
6. Base64-encoded sensitive data

If the user explicitly says "publish anyway despite risks", re-confirm once before proceeding.

---

Placeholder Reference#

Use these placeholders when fixing issues:

TypePlaceholderTokenyour-api-token / <YOUR_TOKEN>API Keyyour-api-key / <API_KEY>Passwordyour-password / <PASSWORD>Emailyou@example.com / <YOUR_EMAIL>Usernameyour-username / <USERNAME>URLhttps://your-server.example.comIPyour-server-ip

---

Scan Script#

The script is at scripts/publish-check.sh relative to this skill's directory. It uses no absolute paths and works in any installation location.

bash

bash scripts/publish-check.sh <target-skill-directory>

---

Safety first, publishing second. Never publish without confirmation. 🛡️