Skip to content
Back to search

pentest-agents

Bug bounty agent framework for Claude Code, Codex, Gemini, Cursor, Windsurf, Copilot, and OpenClaw — 48 agents, 26 commands, 19 CLI tools, 2 MCP servers, autonomous hunt loops, exploit chain builder.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/h-mmer--pentest-agents/badge.svg)](/listings/h-mmer--pentest-agents)

HTML

<a href="/listings/h-mmer--pentest-agents"><img src="/api/listings/h-mmer--pentest-agents/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

# MCP servers are launched via `uv run --with mcp` — no global pip install required.
export HACKERONE_USERNAME=you HACKERONE_TOKEN=your_token
uv run python3 tools/scaffold.py hackerone tesla
cd ~/bounties/hackerone-tesla && claude
/model opus             # Opus 4.7 [1M] — subagents inherit via model: "inherit"
/sync hackerone tesla
/brain init && /status
/hunt tesla.com

`scaffold.py` provisions the workspace for every supported project-scoped client, not only Claude Code: `CLAUDE.md`, `AGENTS.md`, `.codex/`, `.agents/skills/`, `.gemini/`, `.cursor/`, `.windsurf/`, `.github/`, and `.vscode/mcp.json` are generated from the copied workspace assets so paths resolve inside the bounty workspace.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 369 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <p align="center">
  2. <img src="docs/assets/logo.svg" alt="Pentest Agent Suite" width="440"/>
  3. </p>
  4. <h1 align="center">Pentest Agent Suite for Claude Code</h1>
  5. <p align="center">
  6. <em>Autonomous bug-bounty framework for Claude Code and 6 other AI coding tools — 50 agents, 26 commands, 19 CLI tools, 11 skills, 2 MCP servers.</em>
  7. </p>
  8. <p align="center">
  9. <img src="https://img.shields.io/badge/python-3.10%2B-blue?logo=python&logoColor=white" alt="Python 3.10+"/>
  10. <img src="https://img.shields.io/badge/Claude-Code-d97757" alt="Claude Code"/>
  11. <img src="https://img.shields.io/badge/MCP-servers%20%C3%97%202-2ea043" alt="MCP servers"/>
  12. <img src="https://img.shields.io/badge/agents-50-8957e5" alt="50 agents"/>
  13. <img src="https://img.shields.io/badge/payloads-2500%20lines-f85149" alt="Payloads"/>
  14. <img src="https://img.shields.io/badge/IDEs-7-1f6feb" alt="7 IDE targets"/>
  15. </p>
  16. ---
  17. **~760 files · ~118k lines · 50 agents · 26 commands · 19 CLI tools · 11 skills · 2 MCP servers (16 bug-bounty platforms + BYO writeup search) · 2,500 payload lines**
  18. A complete bug bounty framework. Battle-tested hunting methodology with concrete payloads, 7-Question Gate validation, autonomous hunt loops, A→B exploit chain building, persistent brain with endpoint tracking, optional semantic writeup search (bring your own index), automatic cost tracking via CC hooks, live platform integration, and a cross-IDE installer that emits the native format for Claude Code, Codex, Gemini, Cursor, Windsurf, VS Code Copilot, and OpenClaw.
  19. ## Quick Start
  20. ```bash
  21. # MCP servers are launched via `uv run --with mcp` — no global pip install required.
  22. export HACKERONE_USERNAME=you HACKERONE_TOKEN=your_token
  23. uv run python3 tools/scaffold.py hackerone tesla
  24. cd ~/bounties/hackerone-tesla && claude
  25. /model opus # Opus 4.7 [1M] — subagents inherit via model: "inherit"
  26. /sync hackerone tesla
  27. /brain init && /status
  28. /hunt tesla.com
  29. ```
  30. `scaffold.py` provisions the workspace for every supported project-scoped
  31. client, not only Claude Code: `CLAUDE.md`, `AGENTS.md`, `.codex/`,
  32. `.agents/skills/`, `.gemini/`, `.cursor/`, `.windsurf/`, `.github/`, and
  33. `.vscode/mcp.json` are generated from the copied workspace assets so paths
  34. resolve inside the bounty workspace.
  35. ## Install (Claude Code + 6 other AI coding tools)
  36. The framework ships pre-rendered for every supported tool. There are two ways
  37. to use it:
  38. **1. Use the bundles directly (no install step)**
  39. ```bash
  40. git clone https://github.com/H-mmer/pentest-agents-suite
  41. cd pentest-agents-suite/pentest-agents/providers/codex
  42. codex # or: cd ../gemini && gemini, etc.
  43. ```
  44. The `providers/<id>/` tree contains a fully-translated, ready-to-use bundle
  45. for each non-Claude target. Path references inside use `..` to reach the
  46. repo's `tools/`, `rules/`, and `mcp-*-server/` — so the bundle works as
  47. long as it stays inside the cloned repo.
  48. **2. Run the installer (writes into your own project or `~/.codex/` etc.)**
  49. ```bash
  50. python3 -m tools.installer install --targets all --scope project
  51. python3 -m tools.installer install --targets codex --scope global
  52. ```
  53. Install mode rewrites paths to absolute references back into the cloned
  54. pentest-agents repo, so the install works no matter where the user's own
  55. project lives.
  56. | Target | Agents | Slash commands | Rules | MCP | Scopes |
  57. |---|---|---|---|---|---|
  58. | **Claude Code** | native `.claude/agents/*.md` | `.claude/skills/<name>/SKILL.md` | `CLAUDE.md` | `.mcp.json` / `~/.claude.json` | global + project |
  59. | **OpenAI Codex** | native `.codex/agents/*.toml` | `.agents/skills/<name>/SKILL.md` | `AGENTS.md` (≤32 KiB) | `[mcp_servers.*]` in `config.toml` | global + project |
  60. | **Google Gemini** | native `.gemini/agents/*.md` | TOML in `.gemini/commands/` | `GEMINI.md` | `mcpServers` in `settings.json` | global + project |
  61. | **Cursor** | → skills `.cursor/skills/agent-*/SKILL.md` (no native subagents) | → skills `.cursor/skills/cmd-*/SKILL.md` | `.cursor/rules/*.mdc` + `AGENTS.md` | `.cursor/mcp.json` | global + project |
  62. | **Windsurf** | → skills | Workflows | `.windsurf/rules/*.md` (≤12 KiB / file) | `~/.codeium/windsurf/mcp_config.json` | global + project |
  63. | **VS Code Copilot** | `.github/agents/*.agent.md` (≤30 KiB / agent) | `.github/prompts/*.prompt.md` | `.github/copilot-instructions.md` + `.github/instructions/*` | `.vscode/mcp.json` | project + global-MCP |
  64. | **OpenClaw** | → skills | → skills | `~/.openclaw/workspace/AGENTS.md` or `<proj>/AGENTS.md` | `mcp.servers` in `~/.openclaw/openclaw.json` | global + project *(MCP is user-level)* |
  65. Cursor, Windsurf, and OpenClaw have no native subagent concept; Claude-format
  66. agents render as skills/rules. Codex commands are emitted as AgentSkills under
  67. `.agents/skills/`; the deprecated `.codex/prompts/` path is not used.
  68. **`providers/` directory** (in the cloned repo):
  69. ```
  70. providers/
  71. ├── codex/ AGENTS.md + .codex/{agents,config.toml} + .agents/skills
  72. ├── gemini/ GEMINI.md + .gemini/{agents,commands} + settings.json
  73. ├── cursor/ AGENTS.md + .cursor/{rules,skills,mcp.json}
  74. ├── windsurf/ AGENTS.md + .windsurf/{rules,workflows,skills} + mcp_config.json
  75. ├── copilot/ .github/{copilot-instructions.md,instructions,prompts,agents} + .vscode/mcp.json
  76. └── openclaw/ AGENTS.md + .agents/skills/ + openclaw.json
  77. ```
  78. `providers/` is **generated**, not edited by hand. Re-render after editing
  79. `.claude/`, `rules/`, or `skills/` source:
  80. ```bash
  81. python3 -m tools.installer render --targets all
  82. python3 -m tools.installer render --check # exits 1 if drift
  83. ```
  84. The `test_committed_providers_match_render` pytest case enforces drift
  85. detection locally — there is no GitHub Actions CI by project policy.
  86. ### What gets translated
  87. When `.claude/` content is rendered for non-Claude targets, the translator:
  88. - **Drops the `model:` field** — each target uses its own default model.
  89. - **Strips Claude-specific prose** — "Claude Code" → "the AI coding tool",
  90. "the Agent tool" → "the subagent dispatch tool", `model: "inherit"` is
  91. removed entirely.
  92. - **Rewrites `$CLAUDE_PROJECT_DIR`** — to `..` in `providers/` (relative
  93. to the cloned repo), or to absolute paths into the cloned source repo
  94. when installing into a user's project.
  95. - **Maps `effort:` frontmatter** to `model_reasoning_effort` in Codex TOML.
  96. - **Caps body length** — Copilot agents are truncated at 30,000 chars
  97. (Copilot's hard limit). Windsurf rules are chunked at 12,000 chars
  98. (workspace) / 6,000 chars (global).
  99. - **Adds Copilot subagent links** — orchestrator agents (chain-builder,
  100. correlator, recon-ranker) get an `agents:` list of siblings so Copilot
  101. wires the dispatch graph.
  102. ### Installer management
  103. ```bash
  104. pentest-agents list # detect which targets are installed
  105. pentest-agents install --targets claude_code,codex --scope global
  106. pentest-agents install --dry-run # preview every file + JSON merge
  107. pentest-agents verify # check manifest vs. disk (drift)
  108. pentest-agents uninstall # reverse, restore .pa-backup files
  109. pentest-agents render --targets all # regenerate providers/<id>/
  110. pentest-agents render --check # drift gate (exit 1 if dirty)
  111. ```
  112. Every install records a manifest (`.pentest-agents/manifest.json` for project
  113. scope, `~/.config/pentest-agents/manifest.json` for global). Uninstall only
  114. removes files we wrote and surgically strips only the MCP/JSON keys we merged —
  115. your other settings are never touched. Conflicting writes back up the original
  116. as `<path>.pa-backup` and are restored on uninstall.
  117. ## Workflow
  118. ```
  119. New program: /new → /sync → /brain init → /analyze → /surface → /hunt
  120. Returning: /resume <target> → /hunt or /autopilot
  121. After finding: /validate → /chain → /report → /dupcheck → /submit → /learn
  122. Batch triage: /triage (7-Question Gate on all findings)
  123. ```
  124. ## MCP Servers (2)
  125. ### bounty-platforms (16 platforms)
  126. HackerOne (full API), Bugcrowd, Intigriti, Immunefi (public), YesWeHack + 11 stubs.
  127. 7 MCP tools: list_platforms, get_program_scope, get_program_policy, search_hacktivity, sync_program, draft_report, submit_report.
  128. ### writeup-search (BYO index)
  129. Searchable knowledge base agents query during hunting and validation.
  130. 4 MCP tools:
  131. - `search_writeups` — semantic search (FAISS) or keyword search for prior art
  132. - `get_writeup` — full writeup content by ID
  133. - `search_techniques` — exploitation techniques by vuln class
  134. - `search_payloads` — curated payloads from `rules/payloads.md`
  135. > **The writeup index is not bundled.** Bulk-redistributing scraped hacktivity violates most platform ToS, so this repo ships the server only. The `search_payloads` + `search_techniques` fallback works out of the box; the semantic/keyword layers activate once you point the server at your own index.
  136. **Three search modes** (auto-detected, graceful fallback):
  137. | Mode | Requires | Searches |
  138. |------|----------|----------|
  139. | **FAISS** (semantic) | `faiss-cpu`, `sentence-transformers`, your `metadata.db` + `index.faiss` | Your writeup corpus via vector embeddings |
  140. | **SQLite** (keyword) | Your `metadata.db` only | Your writeup corpus via `LIKE` over the text column |
  141. | **Local** (default) | Nothing — zero deps | `rules/payloads.md` + `skills/` shipped in this repo |
  142. Point the server at your index by dropping `metadata.db` (+ optionally `index.faiss`) into `~/.local/share/pentest-writeups/`, or set `WRITEUP_DB_DIR=/path/to/dir`.
  143. **Expected schema** (`metadata.db`): a SQLite file with at least one table containing columns `id`, `title`, `url`, and one text column (`content` / `text` / `body` / `writeup`). Row order in the table must match vector order in `index.faiss` when using semantic mode.
  144. ### Build your own index — `rag-builder/`
  145. The repo now ships a local RAG/FAISS builder under [`rag-builder/`](rag-builder/) that turns a list of GitHub / GitLab repositories into a `metadata.db` + `index.faiss` pair the writeup-search MCP server consumes. Destructive operations (clone, embed, write) are **always gated behind `--execute`** — running the CLI without it prints the plan and changes nothing, so you can never wipe an existing index by accident.
  146. ```bash
  147. cd rag-builder
  148. # 1. Inspect the plan — no network, no writes.
  149. python3 build.py status
  150. python3 build.py ingest # dry-run (the default)
  151. # 2. Opt-in pre-flight: probe every URL with `git ls-remote` (network).
  152. python3 build.py ingest --check-remotes # ~5s for 141 repos at 16 workers
  153. # 3. Actually clone + index every repo from repos.yaml into ./data/.
  154. python3 build.py ingest --execute
  155. python3 build.py ingest --execute --check-remotes # skip unreachable first
  156. # 4. Point the MCP server at the output.
  157. export WRITEUP_DB_DIR="$PWD/data"
  158. python3 ../mcp-writeup-server/server.py --test
  159. ```
  160. `rag-builder/repos.yaml` ships with a 146-entry seed covering CTF archives, bug-bounty reports, payload collections, and research aggregators — edit freely. `repos-skipped.yaml` is loaded automatically as an exclusion list (override with `--skip-list` or `--no-skip-list`). `config.yaml` controls the embedding model (`all-MiniLM-L6-v2` by default), host allowlist, clone size cap, and file-size ceiling. See [`rag-builder/README.md`](rag-builder/README.md) for the full reference.
  161. ## CC Hooks (automatic cost tracking)
  162. Configured in settings.json, fires automatically:
  163. - **SubagentStop** → `cost_hook.py` logs agent name + session to `cost-tracking.json`
  164. - **Stop** → logs session end
  165. - **SessionStart** → welcome message
  166. Statusline shows live cost from session token data: `$0.57`
  167. ## Commands (26)
  168. ### Hunting & Analysis
  169. | Command | Description |
  170. |---------|-------------|
  171. | `/hunt <target> [--vuln-class]` | Active hunting — searches writeup DB for techniques first, then tests with concrete payloads |
  172. | `/autopilot <target>` | Autonomous loop with --paranoid/--normal/--yolo checkpoints |
  173. | `/surface <target>` | P1/P2/Kill ranked attack surface |
  174. | `/chain` | Build A→B→C exploit chains via chain-builder agent (9 capability rows + 4 documented deep chains in `rules/chain-table.md`) |
  175. | `/analyze <target>` | AI analysis: crown jewels, attack paths, blind spots |
  176. | `/mindmap <target>` | Attack surface tree with brain status |
  177. | `/sast <repo>` | Source-code vulnerability hunting (entry → flow → gap → exploit pipeline) |
  178. ### Validation & Reporting
  179. | Command | Description |
  180. |---------|-------------|
  181. | `/validate <finding>` | 7-Question Gate → PASS/KILL/DOWNGRADE/CHAIN REQUIRED |
  182. | `/triage` | Batch-validate ALL findings, kill weak ones |
  183. | `/quality <draft>` | Score report 1-10 (blocks below 7) |
  184. | `/report [format]` | Reports (hard gate: requires /validate PASS) |
  185. | `/dupcheck <desc>` | Hacktivity + writeup DB for duplicates |
  186. | `/submit <finding>` | Submit (hard gate: /validate PASS + /quality ≥ 7) |
  187. ### Session & Memory
  188. | Command | Description |
  189. |---------|-------------|
  190. | `/resume <target>` | Resume — untested endpoints + suggestions |
  191. | `/remember` | Log finding/pattern for cross-target learning |
  192. | `/learn <id> <status>` | Record response — auto-boosts paid techniques |
  193. | `/brain` | init, brief, status, endpoint, endpoints, record, exhausted |
  194. ### Infrastructure
  195. | Command | Description |
  196. |---------|-------------|
  197. | `/new`, `/sync`, `/status` | Setup + dashboard |
  198. | `/pipeline`, `/quickscan`, `/fullscan` | Scanning pipelines |
  199. | `/correlate` | Chain discovery across findings |
  200. | `/cost`, `/monitor` | Cost tracking, target change detection |
  201. ## Agents (50)
  202. ### H1 Weakness Specialists (19)
  203. xss-hunter (#60/#61/#62), sqli-hunter (#67), csrf-hunter (#57), ssrf-hunter (#75), ssti-hunter (#74), idor-hunter (#55), auth-tester (#27), info-disclosure (#18), open-redirect (#38), rce-hunter (#70), xxe-hunter (#63), file-upload (#39), cors-hunter (#58), subdomain-takeover (#145), business-logic (#28), race-condition (#29), privilege-escalation (#26), oauth-hunter (#1/#22/#106/#137), llm-ai-hunter (chains under #18/#55/#61/#70/#106)
  204. ### Hunting & Analysis (3)
  205. - **validator** — 7-Question Gate + never-submit list (PASS/KILL/DOWNGRADE/CHAIN)
  206. - **chain-builder** — A→B chain walk against the capability table, searches writeup DB for proven chains
  207. - **recon-ranker** — P1/P2/Kill surface ranking
  208. ### Infrastructure / Recon (10)
  209. recon, vuln-scanner, config-auditor, cloud-recon, js-analyzer, waf-profiler, graphql-audit, nuclei-writer, browser-agent (Burp MCP), browser-stealth-agent (Camoufox)
  210. ### Meta / Validation (9)
  211. brain, correlator, quality-check, monitor, poc-builder, report-writer, scope-check, browser-verifier (client-side PoC proof), dast-devils-advocate (adversarial downgrade)
  212. ### SAST Pipeline (8)
  213. sast-file-ranker, sast-entry-mapper, sast-danger-mapper, sast-flow-tracer, sast-gap-analyzer, sast-devils-advocate, sast-hunter, sast-exploit-builder
  214. ### Specialized (1)
  215. web3-auditor — Solidity grep arsenal, Foundry PoC, DeFi patterns
  216. ## Hunting Skills (5 deep methodology skills + 6 reference skills = 11)
  217. The hunt-* skills are vuln-class-specific methodology files distilled from
  218. public bug-bounty reports. Each has a verified 2024-2026 CVE catalog and
  219. sub-techniques. The matching specialist agent reads its skill via
  220. `Read $CLAUDE_PROJECT_DIR/skills/hunt-<class>/SKILL.md` before testing.
  221. | Skill | Lines | Pairs With | Highlights |
  222. |-------|-------|------------|------------|
  223. | `skills/hunt-rce/SKILL.md` | 1,135 | rce-hunter | 1,218-report distillation. RSC CVE-2025-55182, runc Leaky Vessels, BentoML pickle, LangChain REPL, Tekton/OpenProject git arg injection, ingress-nginx, container/runtime, ML serving, agentic LLM tool-use, OSS supply chain |
  224. | `skills/hunt-idor/SKILL.md` | 969 | idor-hunter | 1,117-report distillation. Sam Curry automotive chain, OneUptime CVE-2026-30956, Zitadel V2Beta/Mgmt API, Inforcer tenant enum, Apache Answer UUIDv1 prediction, Indico BOLA, GraphQL field-level pivots, agentic AI cross-tenant |
  225. | `skills/hunt-xss/SKILL.md` | 968 | xss-hunter | DOMPurify mXSS family, Auth0 nextjs-auth0 returnTo, RSC DoS family, markdown-to-jsx, listmonk admin-ATO, Trix rich-text editor (H1 #2819573 / #2521419), Jupyter notebook XSS (GHSA-rch3-82jr-f9w9), n8n MCP OAuth XSS (GHSA-537j-gqpc-p7fq), LinkedIn-class iframe-in-article (H1 #2212950), 10 sub-techniques (A-J), Semgrep / ast-grep / ripgrep / CodeQL patterns |
  226. | `skills/hunt-oauth/SKILL.md` | 770 | oauth-hunter | 365-report distillation. ruby-saml parser differentials, Authentik regex `redirect_uri`, workers-oauth-provider PKCE downgrade, Entra ID actor token, Hono JWT alg confusion, nOAuth, Tekton token exfil, Argo CD project token, tinyauth |
  227. | `skills/hunt-llm-ai/SKILL.md` | 930 | llm-ai-hunter | OWASP LLM Top 10 v2025 + Agentic AI Top 10. Microsoft 365 Copilot ASCII Smuggling, LangChain GmailToolkit indirect injection (CVE-2025-46059), LangChain PythonREPLTool semantic RCE (CVE-2025-68613), BentoML pickle, Ollama RCE family, Open WebUI SSE injection, MLflow path traversal |
  228. Reference skills (read by methodology-aware agents): `hunting-methodology`,
  229. `recon-methodology`, `report-writing`, `sast-methodology`,
  230. `triage-validation`, `vuln-classes`.
  231. ## CLI Tools (19)
  232. | Tool | Purpose |
  233. |------|---------|
  234. | brain.py | Brain with endpoint tracking + circuit breaker |
  235. | intel_engine.py | Hacktivity patterns + tech→vuln mapping |
  236. | journal.py | JSONL session journal for /resume |
  237. | target_selector.py | Program ROI ranking |
  238. | cost_hook.py | CC hook: auto-logs agent completions via SubagentStop |
  239. | statusline.py | Dashboard (--compact/--watch/--json) |
  240. | scope_check.py | Scope validation with --list |
  241. | scope_hook.py | PreToolUse hook: blocks out-of-scope Bash commands (exact + wildcard) |
  242. | cvss_version_guard.py | Enforces H1 = CVSS 3.1, other platforms = CVSS 4.0 |
  243. | file_path_guard.py | Blocks hallucinated file paths in reports |
  244. | file_safety.py | Shared safety checks for agent-written files |
  245. | dedup_findings.py | Dedup + hacktivity cross-reference |
  246. | global_brain.py | Cross-engagement knowledge (incremental hash-based sync) |
  247. | response_tracker.py | Response learning + auto-boost paid techniques |
  248. | scaffold.py | Workspace scaffolding with update mode |
  249. | capture.py | Screenshots + video (WSL2) |
  250. | cost.py | Token cost tracking + ROI |
  251. | camofox_ctl.sh | Camoufox (stealth Firefox) lifecycle — Cloudflare/Akamai bypass |
  252. | pentest-statusline.sh | CC statusline: findings, brain, context, cost |
  253. ## Rules Library (`rules/`)
  254. Single source of truth for every agent — all hunters, validators, and report-writers
  255. read the relevant files at session start.
  256. | File | Lines | Purpose |
  257. |------|-------|---------|
  258. | `hunting.md` | 360 | 31 hunting rules (Rule 0 harm check, Rule 8 sibling check, Rule 9 A→B signal, Rule 19 never-submit, Rule 24 mutation matrix, Rule 28 detection-token rotation, Rule 30 no cross-region inference, Rule 31 unauth state-change battery) |
  259. | `payloads.md` | 2,605 | XSS (incl. Detection Mechanism Rotation Ladder) / SSRF / SQLi / IDOR / OAuth / upload / race / SSTI / deser / JWT / LFI / prototype pollution / NoSQLi / DeFi |
  260. | `techniques.md` | 389 | Proven attack techniques extracted from real paid engagements |
  261. | `waf-bypass-protocol.md` | 166 | WAF bypass iteration ladder for Akamai/Cloudflare/Imperva |
  262. | `vendor-status.md` | 127 | Patched vendor vectors, framework fingerprints, cooldown tables |
  263. | `chain-table.md` | 192 | Capability→next-bug chain table for `/chain` (9 capability rows + 4 documented deep chains) |
  264. | `never-submit.md` | 42 | Never-submit list + conditionally-valid-with-chain table |
  265. | `mistakes.md` | 665 | Top 10 most common mistakes — every agent reads this at session start |
  266. ## Key Features
  267. - **Writeup search MCP**: Agents query prior art during hunting — bring your own FAISS/SQLite writeup index, or fall back to the shipped payload/technique library
  268. - **CC hooks**: SubagentStop/Stop auto-log costs, statusline shows live `$X.XX` from token data
  269. - **PreToolUse scope hook**: Bash commands are matched (exact + wildcard) against `scope.yaml`; out-of-scope targets are blocked before the tool call fires
  270. - **7-Question Gate**: Every finding validated — first NO = KILL
  271. - **Depth Engine**: `/autopilot` enforces an anti-shallow protocol — no claim of "exhausted" until the exhaustion matrix is complete
  272. - **Stacked-encoding mandate**: `/hunt` and `/autopilot` require multi-layer encoding in every payload attempt before declaring a surface clean
  273. - **CVSS policy guard**: HackerOne findings use CVSS 3.1; every other platform uses CVSS 4.0 — enforced by `cvss_version_guard.py`
  274. - **Circuit breaker**: 5× consecutive 403/429 → auto-backoff 60s
  275. - **Endpoint tracking**: Brain records every endpoint tested per target
  276. - **Hard validation gates**: /report and /submit refuse without /validate PASS
  277. - **Never-submit filter**: Pipeline auto-kills informational findings
  278. - **Incremental sync**: Global brain hash-based, skips unchanged files
  279. - **Feedback loop**: /learn auto-boosts paid techniques globally
  280. - **Session journal**: JSONL log for /resume continuity
  281. ## Requirements
  282. - Python 3.10+, `uv` (MCP servers launch via `uv run --with mcp`)
  283. - Optional: `uv pip install faiss-cpu sentence-transformers` (for writeup semantic search)
  284. - Security tools: nmap, httpx, subfinder, nuclei, ffuf, katana, sqlmap
  285. - GraphQL hunter tools: `graphql-path-enum` — `cargo install --git https://gitlab.com/dee-see/graphql-path-enum` (auto-installed by `setup-mcp.sh` if `cargo` is present)
  286. - Evidence: grim/scrot, wf-recorder/ffmpeg
  287. - jq (for statusline)
  288. ## License
  289. For authorized security testing only. Follow responsible disclosure.
.claude/settings.json · Bundled script · 173 lines · 2 findings marked
  1. {
  2. "statusLine": {
  3. "type": "command",
  4. "command": "bash $CLAUDE_PROJECT_DIR/tools/pentest-statusline.sh"
  5. },
  6. "hooks": {
  7. "SubagentStop": [
  8. {
  9. "hooks": [
  10. {
  11. "type": "command",
  12. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/cost_hook.py",
  13. "async": true
  14. },
  15. {
  16. "type": "command",
  17. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/chain_pressure_hook.py",
  18. "timeout": 5,
  19. "statusMessage": "Scanning findings for chain-pressure..."
  20. }
  21. ]
  22. }
  23. ],
  24. "Stop": [
  25. {
  26. "hooks": [
  27. {
  28. "type": "command",
  29. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/cost_hook.py",
  30. "async": true
  31. }
  32. ]
  33. }
  34. ],
  35. "SessionStart": [
  36. {
  37. "hooks": [
  38. {
  39. "type": "command",
  40. "command": "echo '{\"hookSpecificOutput\":{\"hookEventName\":\"SessionStart\",\"additionalContext\":\"Bug bounty workspace active. Commands: /status /hunt /surface /resume /triage /chain /autopilot. Check scope FIRST.\"}}'",
  41. "timeout": 5,
  42. "statusMessage": "Loading bug bounty workspace..."
  43. }
  44. ]
  45. }
  46. ],
  47. "PreToolUse": [
  48. {
  49. "matcher": "Bash",
  50. "hooks": [
  51. {
  52. "type": "command",
  53. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/scope_hook.py",
  54. "timeout": 5,
  55. "statusMessage": "Checking scope..."
  56. }
  57. ]
  58. },
  59. {
  60. "matcher": "Write|Edit",
  61. "hooks": [
  62. {
  63. "type": "command",
  64. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/file_path_guard.py",
  65. "timeout": 3,
  66. "statusMessage": "Verifying referenced files..."
  67. },
  68. {
  69. "type": "command",
  70. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/cvss_version_guard.py",
  71. "timeout": 3,
  72. "statusMessage": "Checking CVSS version..."
  73. },
  74. {
  75. "type": "command",
  76. "command": "uv run python3 $CLAUDE_PROJECT_DIR/tools/validity_guard.py",
  77. "timeout": 3,
  78. "statusMessage": "Scanning draft for never-submit signals..."
  79. }
  80. ]
  81. }
  82. ],
  83. "PostToolUse": [
  84. {
  85. "matcher": "Bash",
  86. "hooks": [
  87. {
  88. "type": "command",
  89. "command": "jq -r '.tool_input.command // empty' | { read -r cmd; if echo \"$cmd\" | grep -qE '^(curl|httpx|nuclei|ffuf|nmap|sqlmap|subfinder|katana|feroxbuster|amass|masscan|nikto) '; then uv run python3 \"$CLAUDE_PROJECT_DIR/tools/cost_hook.py\" 2>/dev/null; fi; }",
  90. "timeout": 3,
  91. "async": true
  92. }
  93. ]
  94. }
  95. ]
  96. },
  97. "permissions": {
  98. "allow": [
  99. "Bash(nmap:*)",
  100. "Bash(httpx:*)",
  101. "Bash(subfinder:*)",
  102. "Bash(nuclei:*)",
  103. "Bash(ffuf:*)",
  104. "Bash(katana:*)",
  105. "Bash(curl:*)",
  106. "Bash(uv run python3:*)",
  107. "Bash(python3 $CLAUDE_PROJECT_DIR/tools/*)",
  108. "Bash(python3 tools/*)",
  109. "Bash(dig:*)",
  110. "Bash(whois:*)",
  111. "Bash(host:*)",
  112. "Bash(wafw00f:*)",
  113. "Bash(testssl*)",
  114. "Bash(nikto:*)",
  115. "Bash(wpscan:*)",
  116. "Bash(arjun:*)",
  117. "Bash(gau:*)",
  118. "Bash(waybackurls:*)",
  119. "Bash(subjack:*)",
  120. "Bash(feroxbuster:*)",
  121. "Bash(amass:*)",
  122. "Bash(masscan:*)",
  123. "Bash(rustscan:*)",
  124. "Bash(gowitness:*)",
  125. "Bash(jwt_tool:*)",
  126. "Bash(php:*)",
  127. "Bash(psalm:*)",
  128. "Bash(phpstan:*)",
  129. "Bash(composer:*)",
  130. "Bash(phpggc:*)",
  131. "Bash(semgrep:*)",
  132. "Bash(cppcheck:*)",
  133. "Bash(jq:*)",
  134. "Bash(grep:*)",
  135. "Bash(sort:*)",
  136. "Bash(uniq:*)",
  137. "Bash(wc:*)",
  138. "Bash(cat:*)",
  139. "Bash(head:*)",
  140. "Bash(tail:*)",
  141. "Bash(mkdir:*)",
  142. "Bash(ls:*)",
  143. "Bash(git:*)",
  144. "Bash(gh:*)",
  145. "Read(*)",
  146. "Write(recon/*)",
  147. "Write(scans/*)",
  148. "Write(js-analysis/*)",
  149. "Write(poc/*)",
  150. "Write(reports/*)",
  151. "Write(findings.json)",
  152. "Write(.scope.txt)",
  153. "Write(scope.yaml)",
  154. "Write(policy.md)",
  155. "Write(hacktivity.md)",
  156. "Write(ATTACK_SURFACE_RANKING.md)",
  157. "Write(TRIAGE_REPORT.md)",
  158. "Write(BRAIN.md)",
  159. "Write(cost-tracking.json)",
  160. "Write(sast-work/*)",
  161. "Write(sast-rankings.json)",
  162. "Write(findings/sast/*)",
  163. "Write(poc/sast/*)",
  164. "Edit(*)"
  165. ],
  166. "deny": [
  167. "Bash(rm -rf /)",
    • Dismissed after review: Runs `rm -rf` against a root or home-directory path
    • Line 167 on GitHub
  168. "Bash(rm -rf ~)",
    • Dismissed after review: Runs `rm -rf` against a root or home-directory path
    • Line 168 on GitHub
  169. "Bash(> /dev/sd*)",
  170. "Bash(dd if=*of=/dev/*)"
  171. ]
  172. }
  173. }

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.