| name | fuzzing-apis |
| description | Configure perform API fuzzing to discover edge cases, crashes, and security vulnerabilities.
Use when performing specialized testing.
Trigger with phrases like "fuzz the API", "run fuzzing tests", or "discover edge cases".
|
| allowed-tools | Read, Write, Edit, Grep, Glob, Bash(test:fuzz-*) |
| version | 1.25.0 |
| author | Jeremy Longshore <jeremy@intentsolutions.io> |
| license | MIT |
| tags | ["testing","api","security"] |
| compatibility | Designed for Claude Code, also compatible with Codex and OpenClaw |
API Fuzzer
Overview
Perform API fuzzing to discover crashes, unhandled exceptions, security vulnerabilities, and edge case failures by sending malformed, unexpected, and boundary-value inputs to API endpoints. Supports RESTler (stateful REST API fuzzing), Schemathesis (OpenAPI-driven property-based testing), custom fuzz harnesses with fast-check, and OWASP ZAP active scanning.
Prerequisites
- API specification available (OpenAPI/Swagger, GraphQL SDL, or Protobuf definitions)
- Target API running in a test environment (never fuzz production)
- Fuzzing tool installed (Schemathesis, RESTler, or custom harness with fast-check/Hypothesis)
- API authentication credentials for protected endpoints
- Error logging enabled on the target server to capture crashes and stack traces
Instructions
- Parse the API specification to identify all endpoints, methods, and input schemas:
- Read OpenAPI spec files using Glob (
**/openapi.yaml, **/swagger.json).
- Catalog each endpoint's parameters (path, query, header, body) and their types.
- Note validation constraints (min/max, pattern, enum, required fields).
- Configure the fuzzing strategy:
- Schema-based: Generate inputs that violate schema constraints (wrong types, missing fields, extra fields).
- Mutation-based: Start with valid requests and mutate individual fields (bit flips, boundary values, special characters).
- Dictionary-based: Use known problematic inputs (SQL injection, XSS payloads, format strings, null bytes).
- Define fuzz input categories for each parameter type:
- Strings: Empty, very long (10K+ chars), unicode, null bytes, format strings (
%s%n), path traversal (../../etc/passwd).
- Numbers: 0, -1, MAX_INT, MIN_INT, NaN, Infinity, floats where ints expected.
- Arrays: Empty, single element, thousands of elements, nested arrays, mixed types.
- Objects: Empty, missing required fields, extra unknown fields, deeply nested (100+ levels).
- Dates: Invalid formats, epoch zero, far future, negative timestamps.
- Execute the fuzzing campaign:
- Run Schemathesis:
schemathesis run http://localhost:3000/openapi.json --stateful=links.
- Or run RESTler:
restler-fuzzer fuzz --grammar_file grammar.py.