api-audit

Audits API endpoints across 12 weighted dimensions including validation, auth, and OWASP API security.

6|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/greglas75/zuvo --skill api-audit-greglas75
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: api-audit
Source: https://github.com/greglas75/zuvo/tree/main/skills/api-audit
Command: npx skills add https://github.com/greglas75/zuvo --skill api-audit-greglas75

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? API layers drift over time: endpoints ship without validation, pagination is missing on large tables, auth checks are inconsistent, and contracts diverge between backend and frontend. This Skill runs a structured, scored audit of the entire API surface so teams can find systemic gaps before releases instead of discovering them in production. ## Core Features & Use Cases - 12-dimension scoring (D1-D12): Evaluates input validation, payload efficiency, pagination, error handling, caching, HTTP semantics, client-side waterfalls, rate limiting, auth, documentation, contract stability via oasdiff, and an optional OWASP API Security Top 10 pass (BOLA, BFLA, mass assignment, JWT alg-confusion, GraphQL introspection). - Multi-stack endpoint discovery: Detects and inventories routes across NestJS, Cloudflare Workers, FastAPI, Hono, Next.js, and frontend call patterns (React Query, fetch/axios), cross-checked against OpenAPI specs when present. - Tiered depth and safe probing: LIGHT/STANDARD/DEEP tiers with optional GET-only probing gated by strict environment rules (production refused by default), plus cross-cutting analysis of contract consistency, money fields, and an endpoint-by-role auth matrix. - Use Case: Before a major release of a NestJS backend, run the full audit to score all endpoints, confirm no BOLA vulnerabilities exist on object-level routes, and produce a prioritized findings report with file:line evidence. ## Quick Start Ask the assistant to run the api-audit skill in full mode against the current repository to score every endpoint and generate the audit report.

Frequently Asked Questions about api-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit API endpoints for security and validation issues?

Run the audit in full mode to inventory all endpoints and score them across dimensions D1-D12, including input validation, auth coverage, and OWASP API Security checks like BOLA and mass assignment. Every finding cites a file and line number with severity and a suggested fix.

What does the OWASP API Security Top 10 check (D12) cover?

D12 checks BOLA (object-level authorization), BOPLA excessive data exposure and mass assignment, BFLA function-level authorization, JWT algorithm confusion, sensitive business flow abuse, shadow endpoints, and GraphQL introspection/depth. It activates with the --security flag or automatically when an auth surface is detected.

Does the API audit support NestJS, FastAPI, and Cloudflare Workers?

Yes, stack detection covers NestJS controllers, Cloudflare Workers via wrangler configs, FastAPI routers, Hono, Next.js, and frontend patterns like React Query and raw fetch/axios. Endpoint discovery adapts per stack and cross-checks against OpenAPI specs when present.

Can the audit probe a live API, and is it safe for production?

GET-only probing runs on STANDARD tier and above, but production targets are refused by default; only localhost, staging, or sandbox environments confirmed by the user are probed. Auth tokens must be supplied by the user, and all logged responses are scrubbed of credentials and PII.

What causes an API audit to auto-fail regardless of total score?

Critical gates trigger auto-fail when D1=0 (no input validation), D9<8 (auth gaps on mutations), D3<3 with over 10K rows (unbounded queries), D11=0 (breaking contract change), or any confirmed BOLA, BFLA, or mass-assignment finding on a data-bearing endpoint when D12 is active.

When should I use static-only mode for the API audit?

Use the --static flag when no running server is available or you only want code-level analysis. It skips Phase 2 GET probing entirely while still performing endpoint inventory, dimension scoring, and cross-cutting contract analysis.