hunt-brute-force is a skill for Claude covering missing or weak rate limiting, built from published technique classes. It treats OTP and password-reset brute force, login credential stuffing, username and email enumeration, coupon and gift-card brute force, and ReDoS as one family of findings.
The skill’s central contribution is a four-state classification of rate-limit defenses (hard lockout, soft IP throttle, CAPTCHA injection, and a silent shadow-throttle that a naive test misses), a seeded test that distinguishes a shadow-throttle from an open endpoint, and methods for IP-rotation bypass, timing-based enumeration, and measuring token entropy with a tool. It is for assets covered by a program or engagement you are authorized to test.
What You Get
- A four-state classification of rate-limit defenses, including a seeded test for silent shadow-throttling that a status-code-only check misses
- A phased method covering login rate limits, OTP/2FA brute force, enumeration (string, status and timing), IP-rotation bypass via spoofable headers, token entropy measurement, and ReDoS detection
- Request snippets for ffuf, hydra and nuclei tuned for rate-limit and default-credential testing
- A chain table mapping each finding to its likely severity, from account takeover down to informational
- A validation checklist that requires reachability math for full-keyspace claims and reproducible deltas for enumeration and IP-bypass findings
Common Use Cases
- Testing whether an OTP or 2FA verification endpoint enforces an effective limit across its full numeric keyspace
- Checking whether a login or password-reset endpoint can be rate-limit-bypassed by rotating X-Forwarded-For and similar headers
- Measuring the entropy of a password-reset or session token to see whether it is predictable within its validity window
What Makes It Unique
Most rate-limiting checks stop at whether a 429 appears. This skill’s four-state model names the case where a server returns 200/401 for every request while silently dropping the submissions that would succeed, and gives a seeded known-good probe to detect that case directly. It also ties severity for a brute-force claim to measured throughput against keyspace size, not a short probe.
Under The Hood
The skill is a single SKILL.md of guidance with no bundled scripts. Following it, the agent sends bursts of HTTP requests to login, OTP-verification, password-reset and registration endpoints, measuring status code, response latency and body size across many requests, and may run external tools such as ffuf, hydra, nuclei and Burp Sequencer against those endpoints. It writes intermediate output such as collected tokens to the working folder for entropy analysis. Use it only against assets you are authorized to test.