# h-pentest-ai-platform

h-pentest-ai-platform — a skill that walks an agent through standing up H-Pentest: five LLM roles plan the test, suggest payloads, scan with Nuclei and write up what they find.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode, Windsurf (Devin Desktop), Claude.ai · 248 installs · ★ 13 repo stars
- Author: reason-machines
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/reason-machines/security-skills/tree/HEAD/skills/h-pentest-ai-platform/SKILL.md
- Updated: 2026-10-09
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/skills/h-pentest-ai-platform

## What it does

The skill teaches your agent to stand up and drive H-Pentest, a pentest platform split across five LLM roles. A meta supervisor calls time on the run, a strategic one redraws the plan each round, a worker follows a reason-act-observe loop, a payload role offers vectors every 3 rounds, and a report role turns the transcript into attack paths. Behind them sits a knowledge base of 52+ attack documents covering IDOR, SQL injection, XSS, SSRF, XXE and file upload, fetched through retrieval.

Docker Compose brings the whole thing up, or you can run the pieces yourself: the Python backend needs `pip install -r requirements.txt` and `python init_db.py`, the frontend runs on npm. The API listens on `http://localhost:8000` with docs at `/docs`, the UI on `http://localhost:5173`. Progress arrives over WebSocket, and a message to the intervention endpoint nudges a running agent. Python code runs inside a Docker sandbox alongside Nuclei and a directory scanner.

Both keys live in `config.json`: `openai.api_key` for the models, `dashscope.api_key` for embeddings. Knowledge-base search depends on the second one. There are two modes, `ctf` and `realworld`, and the meta supervisor takes the choice into account. Sandbox limits sit in the same file: `execute_python` gets 30 seconds and 512m of memory, Nuclei runs at critical, high and medium severity.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/security-skills --skill h-pentest-ai-platform -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/h-pentest-ai-platform/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/h-pentest-ai-platform/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/security-skills --skill h-pentest-ai-platform -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/h-pentest-ai-platform/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/security-skills --skill h-pentest-ai-platform -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/h-pentest-ai-platform/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install reason-machines/security-skills skills/h-pentest-ai-platform
```

2. **Manually: put the files in place**
- File: `.github/skills/h-pentest-ai-platform/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/security-skills --skill h-pentest-ai-platform -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/reason-machines/security-skills --path skills/h-pentest-ai-platform
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/h-pentest-ai-platform/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add reason-machines/security-skills --skill h-pentest-ai-platform -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/reason-machines/security-skills/tree/HEAD/skills/h-pentest-ai-platform/SKILL.md

## When to use it
- bring the platform up from a Compose file
- create a `ctf` task against a target URL
- look up a bypass technique in the attack knowledge base
- wire Nuclei into the agent and filter findings by severity
- step into a running agent and point it at another endpoint

## Example prompts
- Set up H-Pentest with Docker Compose and tell me which keys go into `config.json`.
- Create a `ctf` task for `http://target-ctf.example.com` with 25 rounds and follow it over WebSocket.
- Run a `realworld` scan on the staging app and send the worker to the `/api/admin` endpoint.

## Similar
- [pentest-engagement](https://affpapa.org/en/nevibecode/skills/pentest-engagement.md) — skill · 157 installs
- [supabase-pentest](https://affpapa.org/en/nevibecode/skills/supabase-pentest.md) — skill · 748 installs
- [cairn-ai-pentest](https://affpapa.org/en/nevibecode/skills/cairn-ai-pentest.md) — skill · 378 installs
- [pentest-engagement-methodology](https://affpapa.org/en/nevibecode/skills/pentest-engagement-methodology.md) — skill · 356 installs
- [net-pentest](https://affpapa.org/en/nevibecode/skills/net-pentest.md) — skill · ★ 106 repo stars
- [web-pentest](https://affpapa.org/en/nevibecode/skills/web-pentest.md) — skill · 722 installs

From the same set: [reason-machines/security-skills](https://affpapa.org/en/nevibecode/by/reason-machines/security-skills.md) — 129 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
