Security

Trusted ClawHub Install Gate

Try it

Audit a local OpenClaw or ClawHub skill directory before installation, classify it as PASS, REVIEW, or BLOCK, check active-project impact, and optionally run...

What it does

Audit a local OpenClaw or ClawHub skill directory before installation, classify it as PASS, REVIEW, or BLOCK, check active-project impact, and optionally run...

The skill document

Trusted ClawHub Install Gate

Use this skill when a user wants to inspect a local skill artifact before install. The default result is an audit report, not an installation.

Workflow:

  1. Run clawhub-install-gate inspect first.
  2. Explain the findings and classify the artifact as PASS, REVIEW, or BLOCK.
  3. Before any install, perform an active-project impact check:
    • exact destination skills directory,
    • whether the install is workspace-local or global,
    • whether a skill directory with the same name already exists,
    • whether .clawhub/lock.json or another lockfile may change,
    • required binaries, environment variables, accounts, or network access,
    • scripts, hooks, assets, and references that would become active,
    • whether --replace or --allow-review would be needed.
  4. Do not install on BLOCK.
  5. Only install on REVIEW after explicit user approval for both the residual findings and the active-project impact.
  6. Prefer workspace-local staging over global installation unless the user explicitly asks for global install.
  7. Use clawhub-install-gate verify after install when requested.
  8. Use clawhub-install-gate usage when the user asks to track install or REVIEW override usage on this machine.

Active-project decision:

  • NO INSTALL: BLOCK, unclear provenance, unknown destination, duplicate name without replace approval, global destination without explicit request, hidden hooks, credential access, service restarts, or unreviewed scripts.
  • STAGE ONLY: PASS or approved REVIEW, but target runtime, duplicate-name risk, dependency impact, or proof value is still uncertain.
  • INSTALL: PASS, destination is explicit, no duplicate-name surprise, impact is understood, and the user asked to install.

Review policy:

  • Treat the verdict as an Auto-review-style boundary decision, not a permission grant.
  • BLOCK covers private-data transmission, secret-store probing, broad security weakening, denial circumvention, and destructive actions.
  • If a risky action is denied, do not route around it; choose a materially safer path or stop for user decision.
  • Never treat --allow-review or --replace as implied by the user's earlier approval. Confirm those flags for the specific artifact and destination.

Related skills

ClawHub reputation checker for clawsec-suite. Adds a standalone reputation gate before guarded skill installation.

34 installs

Deep audit for installed ClawHub skills — usage analysis, permission review, conflict detection

4 installs

Use before installing, enabling, or running any third-party OpenClaw skill, and when the user says "install this skill", "is this skill safe", "scan/vet/check this skill", or "should I trust this". Also use when a skill is pulled from ClawHub or any untrusted source.

2 installs

Install or upgrade a ClawHub skill, then make the global copy under the machine's OpenClaw home `skills/` directory the final source of truth. Use when the u...

14 installs

Audit a named ClawHub skill or skill URL before installation by combining OpenClaw verification with bounded static analysis. Use when the user explicitly asks whether a skill is safe or requests a pre-install review; report evidence and uncertainty instead of treating a score as proof.

39 installs

Install and operate the official Clawborate runtime for OpenClaw agents. Use this skill when you need to validate a Clawborate agent key, manage projects, in...

19 installs1 stars