agentleFS
Sign inStart free
OpenCovenOpenCoven/coven/skills/openclaw-trust/SKILL.md

openclaw-trust

Skill75 starsChanged 4 months agoMITView on GitHub

Apply OpenClaw’s official threat model and trust guidance (MITRE ATLAS–mapped) when evaluating risky workflows, skills, channel exposure, prompt injection, and tool execution safety.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the user asks about security posture, threat modeling, attack vectors, trust boundaries, or safe rollout decisions for …

    line 8
  2. Distinctive:

    When recommending hardening, prefer specific actions:

    line 31
  3. Very specific:

    Default-deny sender/channel access (AllowFrom least privilege)

    line 33
  4. Distinctive:

    When drafting a new threat proposal, include:

    line 47
  5. Distinctive:

    If any answer is “no/unknown”, treat as medium+ risk and recommend guardrails before rollout.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

OpenClaw Trust

Use this skill when the user asks about security posture, threat modeling, attack vectors, trust boundaries, or safe rollout decisions for OpenClaw automations.

Source of truth

  • Repo: openclaw/trust
  • Core files:
    • threats.yaml — machine-readable threat model
    • CONTRIBUTING.md — how to submit threats/mitigations/chains
  • Public render: trust.openclaw.ai/trust/threatmodel
  • Vulnerability reporting: security@openclaw.ai

What to apply in practice

1) Default to the highest-risk classes first

Prioritize controls for risks repeatedly marked high/critical in the trust model:

  • Prompt injection (direct + indirect)
  • Malicious skills / compromised skill updates
  • Token theft from local config
  • Tool argument injection and command execution abuse
  • MCP/tool over-permission

2) Translate trust model into concrete controls

When recommending hardening, prefer specific actions:

  • Default-deny sender/channel access (AllowFrom least privilege)
  • Approval gates for state-changing actions
  • Strict tool argument validation (no free-form shell interpolation)
  • Limit/monitor skill install/update sources
  • Least-privilege MCP server config
  • Audit logs for new sender authorization + sensitive actions
  • Token hygiene (rotation, minimize plaintext exposure)

3) Separate threat reporting vs vuln disclosure

  • Threat model additions/updates → openclaw/trust issues/PRs
  • Live exploitable vulnerabilities → responsible disclosure via Trust page/security email

Contribution pattern (for writing trust updates)

When drafting a new threat proposal, include:

  • scenario + attack path
  • affected OpenClaw components
  • rough risk estimate
  • suggested mitigation(s)

Maintainers will map to MITRE ATLAS tactic/technique + assign IDs.

Fast mental checklist for risky automation

  • Can untrusted content reach prompt/tool parameters?
  • Is there a hard approval gate before side effects?
  • Could this run with broader permissions than required?
  • Is sender/tool/source authenticated and allowlisted?
  • Are outputs validated before execution/posting/sending?

If any answer is “no/unknown”, treat as medium+ risk and recommend guardrails before rollout.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free