Docs/Skills & MCP tools

flowrail-design-review

live

When Claude reads your spec for the first time, this skill registers a design review with FlowRail and pulls back the predicted threats. Every other check threads through the review id this skill creates.

01

When it runs

Automatically, when Claude is about to write a file matching specs/**/*.spec.md, or when you explicitly ask it to open a review. You don't invoke it directly.

What you see in Claude Codev0.4.2
↳ flowrail-design-review
  spec       specs/password-reset.spec.md
  threats    4 (2 high, 2 med)
  review     dr_7f3a9e2c
  approved   argon2@^0.40, jose@^5.9
  channels   api.stripe.com
MCP toolv0.4.2
flowrail_design_review
Requestv0.4.2
{
  "spec_path":    "specs/password-reset.spec.md",
  "spec_content": "# Password reset flow\n\n...",
  "repo_signals": {
    "language":        "typescript",
    "package_manager": "npm",
    "existing_deps":   ["express", "pg"]
  }
}
Responsev0.4.2
{
  "design_review_id":      "dr_7f3a9e2c",
  "threats":               [{ "category": "...", "severity": "...", "summary": "..." }],
  "approved_dependencies": [{ "name": "...", "version_spec": "..." }],
  "channel_allowlist":     ["api.example.com"]
}
02

Side effect

Writes .flowrail/context.json with active_design_review_id. Server-side, emits design_review_completed scoped to your workspace. Spec text is hashed (spec_content_sha256) and discarded.