hunt-api-misconfig is a skill for Claude that covers API security misconfiguration in authorized testing. Its topics are mass assignment on profile and account endpoints, JWT signature and algorithm problems, server-side prototype pollution, HTTP verb tampering and permissive CORS settings with credentials.
It also covers two areas added for 2024 to 2026: bypasses of filtering in OData query options, and exposed API specification files from NSwag, Swagger, springdoc, FastAPI and Quarkus. For prototype pollution it asks the agent to find an object-merge point first and to prove that the polluted state reaches a later operation before calling it a finding.
What You Get
- Notes on mass assignment, JWT algorithm and signature problems, verb tampering and CORS with credentials
- A stepwise sequence for server-side prototype pollution that demands proof of a downstream effect
- OData query-option filtering bypass notes
- Common paths where Swagger and OpenAPI specs are exposed for several frameworks
- Links to related skills for chains
Common Use Cases
- Testing profile and account update endpoints for fields the server should not accept
- Checking whether a documented API exposes its specification and hidden endpoints
- Reviewing a JWT-based login flow in a bug bounty scope
What Makes It Unique
It is limited to API configuration flaws and sets a proof bar for each: a prototype-pollution claim needs a visible later effect, and a CORS claim needs credentials in the response headers. The framework-specific spec paths make it quick to apply on .NET, Java, Python and Quarkus stacks.
Under The Hood
The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent sends HTTP requests to API endpoints in scope with a command-line client, compares responses against a baseline, and may write notes to the working folder. The text includes short code examples of manipulated tokens and request bodies as teaching material. Use it only against assets you are authorized to test.