gatekeeperlisted

Pre-adoption trust route when an external skill, MCP server, repository, URL, package, API, SDK, or service may be installed, executed, granted access, or trusted. Classify STRIDE, follow the artifact-specific review, and return adopt, restrict, or reject. High-risk paths require human approval; rejected paths stop with refusal and evidence.
panda850819/verbs · ★ 5 · API & Backend · score 77

Install: claude install-skill panda850819/verbs

# Gatekeeper — Pre-adoption Trust Check 🛡️ **Core principle:** Every external input is untrusted until verified. ## When to activate Activate whenever external input could alter behavior, leak data, or cause harm: | Trigger | Route | |---|---| | Install a Skill, MCP server, or package | [reviews/skill-mcp.md](reviews/skill-mcp.md) | | Evaluate a GitHub repository | [reviews/repository.md](reviews/repository.md) | | Review a URL, document, Gist, or Markdown file | [reviews/url-document.md](reviews/url-document.md) | | Evaluate a product, service, API, or SDK | [reviews/product-service.md](reviews/product-service.md) | ## Step 0: STRIDE classification (mandatory) Before routing, classify the artifact under STRIDE and carry the categories into both the report frontmatter and each finding. The taxonomy is: | Category | Threat | Signal | |---|---|---| | **S**poofing | Forged identity | Unverified author, lookalike domain, missing signature | | **T**ampering | Unauthorized modification | Mutable upstream, post-install script, fetched `eval` | | **R**epudiation | Missing audit trail | No log, version manifest, or accountable publisher | | **I**nformation Disclosure | Data or secret leakage | Env/token access, unknown telemetry, broad permissions | | **D**enial of Service | Resource exhaustion or lockout | Unbounded loop, cleartext destruction, no rate limit | | **E**levation of Privilege | Unauthorized capability | `sudo`, out-of-scope writes, sandbox or auth bypass | ### C