metaphor/marketplace

skill · built into metaphor

/ai-security

Budgeted security and AI-agent risk review — ai_security scan, OSV CVE/GHSA lookups, OWASP LLM/MCP/Top10 tagging, draft for Agent Review canvas

  • Built in
  • Slash command
try it

you ›/ai-security

metaphor ›Loaded the ai-security skill. What should I work on?

What the agent is told

This is the skill's SKILL.md, exactly what Metaphor reads before the task.

Use when the ask is a security review, CVE/dependency audit, OWASP mapping, or AI/agent risk (skills, MCP tool poisoning, prompt injection) — not a CI-only ask and not a full multi-persona PR review.

Prefer to delegate: task({ subagent_type: "security", description, prompt }) (background) → await_subagent → call pr_review with the child's pr_review_draft so the Agent Review canvas opens.

Structured Agent Review (/review @pr:N / CTA): when pr-review-security is active, spawn security in the same first turn as pr (they run in parallel). Do not run the scan on the primary.

Tools

Tool When
ai_security({ action: "scan", path|paths|text, focus? }) First pass on skills, MCP configs, prompts, or code
ai_security({ action: "catalog", framework? }) Offline LLM/MCP/Top10/agentic id maps
osv_query Package+version, batch, or CVE/GHSA/MAL id
owasp_lookup Deep-dive one category/topic (optional cheat-sheet fetch)
bash pnpm audit / npm audit; optional gh api …/dependabot/alerts (soft-fail 403)

Ladder

  1. Scope — pin PR / paths / skill / MCP config.
  2. Map — diff or local tree; list AI surfaces.
  3. Scan — ai_security scan those surfaces (includeDeps when manifests change).
  4. Static appsec — authz, secrets, injection with evidence loci.
  5. Deps — audit + osv_query for hits in this tree only.
  6. Tag — OWASP/LLM/MCP ids on real findings only.
  7. Conclude — Verdict + pr_review_draft for the parent.

Output

Parent calls pr_review once. Chat stays to one line. Nothing posts to GitHub until a human publishes from the canvas.

Agents that use it

Questions

What is the ai-security skill?

Budgeted security and AI-agent risk review — ai_security scan, OSV CVE/GHSA lookups, OWASP LLM/MCP/Top10 tagging, draft for Agent Review canvas

How do I use ai-security in Metaphor?

It is built into Metaphor. Type /ai-security in the composer, or describe the task and the agent loads the skill when it fits.

What is a skill?

A skill is a set of written instructions the agent reads before it starts a kind of task. It changes how the agent works, not what it is allowed to do.

More built-in skills