hunt-race-condition
Skill0 starsChanged 5 months agoMITView on GitHub
Hunting playbook for race-condition / TOCTOU vulnerabilities — concurrent-request abuse, HTTP/2 single-packet attacks, double-redeem / double-spend, MFA-OTP & email-verify races, rate-limit bypass. Use when hunting races or TOCTOU bugs.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- murrtada/bug-bounty-agent-skillstheir own version, first seen Nov 15, 2026
- ajtazer/heckittheir own version, first seen Sep 11, 2026
- elementalsouls/Claude-BugHuntertheir own version, first seen Jul 1, 2026
- hhjkjkjk/Claude-skillstheir own version, first seen May 29, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- elementalsouls/Claude-BugHunter★ 6.1ksame name
- murrtada/bug-bounty-agent-skills★ 7same name
- ajtazer/heckit★ 2same name
- hhjkjkjk/Claude-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository63 other files it gives its agents.
- .claude/skills/nestjs-best-practices/AGENTS.mdAGENTS.md
- accessibility-a11ySkill
- architecture-blueprint-generatorSkill
- backend-delivery-loopSkill
- backend-designSkill
- backend-dev-guidelinesSkill
- backend-testingSkill
- backend-to-frontend-handoff-docsSkill
- building-componentsSkill
- bullmq-specialistSkill
- canvas-designSkill
- code-patternsSkill
- components-composition-patternsSkill
- conventional-commitSkill
- core-web-vitalsSkill
- create-svgSkill
- css-architectureSkill
- cypress-authorSkill
- cypress-docsSkill
- cypress-explainSkill
- delivery-loop-retroSkill
- diagnoseSkill
- fastify-best-practicesSkill
- figma-code-connectSkill
- frontend-delivery-loopSkill
- frontend-fundamentalsSkill
- fullstack-delivery-loopSkill
- grill-meSkill
- impeccableSkill
- mermaid-diagramsSkill
- nestjs-best-practicesSkill
- nextjs-best-practicesSkill
- nextjs-performanceSkill
- nodejs-backend-patternsSkill
- owasp-security-reviewSkill
- performance-hunterSkill
- playwright-best-practicesSkill
- postgresql-code-reviewSkill
- postgresql-optimizationSkill
- postmanSkill
- prisma-postgresSkill
- pro-state-machineSkill
- race-conditionSkill
- react-best-practicesSkill
- react-performanceSkill
- react-querySkill
- redis-coreSkill
- redis-query-engineSkill
- redis-securitySkill
- refactorSkill
- review-panelSkill
- run-delivery-chainSkill
- spec-drivenSkill
- sql-optimizationSkill
- system-designSkill
- testing-strategySkill
- tune-skills-and-agentsSkill
- ui-ux-pro-maxSkill
- vercel-react-view-transitionsSkill
- viteSkill
- The first 60 shown.
The file
Files
Hunting Race Conditions
A hunting playbook for race-condition / TOCTOU vulnerabilities, built from 12 public bug-bounty and coordinated-disclosure cases. Race conditions are high-severity because they break the financial, access-control, and integrity assumptions defenders rarely stress-test: double-spending credits, inflating votes, bypassing per-user limits, and racing token/state transitions. The enabling primitive for modern targets is the HTTP/2 single-packet attack, which collapses the race window to sub-millisecond and makes rate-limited, distributed, load-balanced backends race-able.
Quick-start
- Enumerate one-time / limited-use actions — every endpoint enforcing "once per user", "limited quantity", or "deduct balance". The gap between its state read and state write is your window.
- Capture a clean baseline request and confirm the expected single-use behaviour.
- Stage parallel requests — prefer Turbo Intruder
Engine.BURP2(single-packet HTTP/2) when the target advertisesh2; otherwise fall back to parallel HTTP/1.1. - Fire 10–50 simultaneous requests with last-byte sync, then look for multiple successes, duplicate effects, or stale-state where only one should have won.
- Verify the effect and reproducibility (3/5 attempts), then run Gate 0 before reporting.
Full detail is loaded on demand — see References.
References
Each file is loaded on demand — read one only when the task needs that depth (progressive disclosure).
references/methodology.md— crown-jewel targets, attack-surface signals (URLs, headers, JS/tech-stack tells), the 10-step hunting methodology, common root causes, defender bypass techniques, Gate 0 validation, and real impact examples · read when scoping targets, planning the hunt, or validating/writing up a finding.references/payloads.md— copy-paste Turbo Intruder,curl-parallel, and Pythonasynciorace templates, plus source-audit grep patterns and the HTTP/2 capability check · read when you need a ready-to-fire PoC or to audit source for unlocked read-then-write.references/single-packet-attack.md— the HTTP/2 single-packet deep reference: why it works, last-byte-sync mechanics, Wireshark validation, h2.0 vs h2.cl variants, race-window estimation, connection/stream shapes, the annotatedEngine.BURP2template, Flatt's first-sequence-sync (N>30), operator playbook, and anti-patterns · read when building or debugging a single-packet exploit or deciding which attack shape to use.references/disclosed-reports.md— the per-bug-class catalog of the 12 disclosed cases (GitLab CVE-2022-4037, Worldcoin, Stripe ×2, Reverb gift card, Cosmos faucet, InnoGames, Flatt PIN-bruteforce, nopCommerce CVE-2024-58248, …) with payload, root cause, and bounty · read when matching your target to a known race class or citing precedent.
Related Skills & Chains
hunt-business-logic— Race conditions are the "concurrency arm" of every business-logic state machine. Chain primitive: business logic (coupon/promo) + race-condition single-packet attack → coupon redeemed N times → direct financial loss.hunt-mfa-bypass— OTP-expiry windows and replay protection are classic race targets. Chain primitive: race + MFA-validate endpoint → bypass OTP expiry by submitting N concurrent validations within the validity window.hunt-ato— Race conditions on password reset, email change, and account creation enable persistent ATO. Chain primitive: race on email-change endpoint + atomic-update missing → swap victim email + read reset token before user notice.hunt-api-misconfig— Wallet/balance/credit endpoints without atomic UPDATE are double-spend candidates. Chain primitive: race + atomic-update missing → double-spend balance → withdraw N× user balance.security-arsenal— Load the Turbo Intruder single-packet template, h2.cl smuggling for atomic submit, andcurl --nextparallel multi-request patterns.triage-validation— Apply the Statistical-Sampling gate: a single anomalous response is noise; require 1 successful + N duplicate / over-quota / stale-state demonstrations with response screenshots before reporting.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

