Sentinel

Universal AI agent security scanner that orchestrates a team of 6 specialist security agents to produce comprehensive security assessments.

Install

npx create-sentinel-security

The installer prompts for a target platform, then copies the correct Sentinel files into your current project.

Supported Platforms

1. GitHub Copilot
2. Claude Code
3. Cursor
4. Windsurf
5. Codex
6. Cline
7. Aider
8. Gemini CLI
9. Crush (OpenCode successor)
10. Roo Code
11. Amazon Q Developer
12. Amp
13. Continue.dev

Installed Paths By Platform

| Platform | Installed Files | | ---------- | ------------------ | | GitHub Copilot | .github/agents/*.agent.md | | Claude Code | .claude/agents/*.md | | Cursor | .cursor/rules/*.mdc | | Windsurf | .windsurf/rules/*.md | | Codex | AGENTS.md and .sentinel/*.md | | Cline | .clinerules/*.md | | Aider | CONVENTIONS.md and .sentinel/*.md | | Gemini CLI | GEMINI.md and .sentinel/*.md | | Crush | AGENTS.md and .sentinel/*.md | | Roo Code | .roo/rules/*.md | | Amazon Q Developer | .amazonq/rules/*.md | | Amp | AGENTS.md and .sentinel/*.md | | Continue.dev | .continue/rules/*.md |

What It Does

Sentinel scans AI agent definitions, rules, skills, and instruction files for security vulnerabilities. It works with ANY AI tool — Claude Code, Cursor, Windsurf, GitHub Copilot, OpenClaw, Aider, Gemini, Crush, Roo Code, Amazon Q, Amp, Continue, GitLab Duo, MCP servers, and more. Each scan produces a graded security assessment with actionable remediation guidance.

Usage

Invoke Sentinel with a target:

• A URL (GitHub repo, raw file, agent marketplace, etc.)
• A local file path
• Pasted agent definition text

Platform examples:

• GitHub Copilot / Claude Code / Cursor / Windsurf: use @sentinel
• Codex: put instructions in AGENTS.md and run Codex tasks normally
• Cline: keep Sentinel in .clinerules/ and invoke in Cline chat
• Aider: keep Sentinel in CONVENTIONS.md and .sentinel/ specialist files
• Gemini CLI: keep Sentinel in GEMINI.md and import .sentinel/ specialists
• Crush: keep Sentinel in AGENTS.md and .sentinel/ specialist files
• Roo Code: keep Sentinel in .roo/rules/ files
• Amazon Q Developer: keep Sentinel in .amazonq/rules/ files
• Amp: keep Sentinel in AGENTS.md and @-mention .sentinel/ specialists
• Continue.dev: keep Sentinel in .continue/rules/ files

The Team

| Agent | Role | Focus | | ------- | ------ | ------- | | Red Hat | Offensive security | Attack vectors, exploitation methods, proof-of-concept payloads | | Blue Hat | Defensive architecture | Protections, sandboxing, permission scoping, credential management | | White Hat | Standards & compliance | OWASP Top 10 for LLM Applications audit | | Yellow Hat | Prompt injection | Injection, manipulation, and jailbreak resistance | | Green Hat | Remediation | Hardened agent definitions, Agent Constitution, CI rules | | Purple Hat | Synthesis | Cross-reference findings, validate fixes, final grade |

Grading Scale

• A (90-100): No critical issues, minimal warnings — safe to deploy
• B (75-89): No critical issues, some warnings — deploy with noted caveats
• C (50-74): 1-2 critical issues, multiple warnings — fix before deploying
• D (25-49): Multiple critical issues, significant gaps — major rework needed
• F (0-24): Dangerous — do not deploy without complete overhaul

Test Agents

See test-agents/ for example input (vulnerable agent) and output (hardened agent + remediation report).

License

MIT