NiceFox Secu

AI-powered security review for web developers. Find vulnerabilities, fix them in your code, verify the fixes — all in one session.

Zero security experience needed. Works with any AI coding agent: Claude Code, Codex, opencode, Cursor, Kimi, aider...

Based on the AIDA methodology.

Quick Start

1. Run NiceFox Secu

From your project directory:

npx nicefox-secu

On first run, this builds the security toolkit (~1GB Docker image, takes ~2-3 min once).

2. Start the Security Review

Open your AI coding agent and paste:

Read ~/.nicefox/REVIEW.md and start the security review

The AI will:

1. Auto-detect your framework, target URL, and environment (dev/prod)
2. Ask you to confirm — one question instead of a setup wizard
3. Review your app (recon, endpoint mapping, vulnerability testing)
4. Fix each vulnerability directly in your source code
5. Verify each fix by re-testing
6. Print a summary of what was found and fixed

What You Need

• Docker — https://docs.docker.com/get-docker/
• An AI coding agent — Claude Code, Codex, opencode, Cursor, Kimi, aider...

That's it.

Example

$ cd ~/projects/my-express-api
$ npx nicefox-secu

  NiceFox Secu
  AI-powered security review for web developers

  ✓ Docker is running
  ✓ Security toolkit image ready
  ✓ Toolkit container running
  ✓ Prompt installed

  Ready! Open your AI coding agent from your project directory and paste:

    Read ~/.nicefox/REVIEW.md and start the security review

$ claude  # or opencode, cursor, codex, aider...

> Read ~/.nicefox/REVIEW.md and start the security review

AI: Detected: Express.js project, target http://localhost:3000, dev mode.
    Start the security review? (Y/n)

You: Y

AI: [scanning, testing, fixing...]

AI: Found VULN-001: SQL Injection in POST /api/search (CRITICAL)
    Fixing src/routes/search.js — parameterized query...
    Fix verified.

AI: Found VULN-002: Missing rate limiting on POST /api/login (MEDIUM)
    Fixing src/routes/auth.js — adding express-rate-limit...
    Fix verified.

AI: Assessment complete.
    2 vulnerabilities found, 2 fixed, 0 require manual attention.

Environment Modes

Development (auto-detected when target is localhost):

• Aggressive scanning, all exploitation techniques allowed
• AI edits your source code directly to fix vulnerabilities
• Full tool suite

Production — pass the target URL as argument:

npx nicefox-secu https://example.com

• Non-destructive tests only, rate limits respected, extra caution on risky tests
• If source code is available locally, AI reads it and applies fixes; otherwise documents recommended fixes only
• Asks for authorization confirmation before scanning
• Asks which paths/subdomains to exclude

Included Tools

The security toolkit Docker image ships with:

Category	Tools
Recon	nmap, subfinder
Vuln scanning	nuclei
Web discovery	ffuf
Parameters	arjun
SQL injection	sqlmap
XSS	dalfox
API testing	httpie, curl
JWT	jwt_tool
Brute force	hydra
Wordlists	SecLists (Discovery, Fuzzing, Passwords)

Safety

• Always test against a dev/staging environment first
• Never test production systems without explicit authorization
• Backup your code before running (or just use git — you do use git, right?)
• In prod mode, the AI applies fixes locally if source code is available, or documents them otherwise

Troubleshooting

Rebuild the toolkit image

docker rm -f nicefox-tools        # Remove container
docker rmi nicefox-tools           # Remove image
npx nicefox-secu                   # Rebuilds from scratch

Tools not responding

docker exec nicefox-tools nmap --version    # Test a tool manually

macOS / Windows networking

If tools can't reach your local app, the AI will automatically use host.docker.internal instead of localhost — this is handled in the prompt.

License

MIT — Use at your own risk. Only test systems you own or have explicit permission to test.

Credits

Based on AIDA (AI-Driven Security Assessment), simplified for web developers.