nullsec-mcp v0.1.0
Trust analysis for MCP servers and AI agents.
Features
- Tool Inventory — discovers all registered MCP tools, classifies capability types (filesystem, shell, network, database, wallet, credentials)
- Dangerous Capability Detection — flags tools with unguarded dangerous operations; distinguishes mitigated vs unguarded
- Permission Analysis — maps credential surface from env vars, README docs, and source code
- Injection Surface — AI-powered detection of confused-deputy attacks, unsanitized outputs, and tool-chaining risks (requires
ANTHROPIC_API_KEY)
- Input Validation — static analysis for path traversal, SQL injection, command injection, SSRF, eval
- Network Egress — identifies outbound domains and dynamic URL fetching
- Trust Score — deterministic 0–100 score with grade (A+ to F), risk level, and capability breakdown
- NSIP Manifest — generates machine-readable
nsip.json trust documents (nsip/v0.1 schema)
Usage
npx nullsec-mcp https://github.com/org/your-mcp-server
Options
| Flag |
Description |
--json |
Full JSON output for CI/CD |
--nsip |
Write nsip.json to current directory |
--no-ai |
Skip AI analysis (no API key needed) |
--max-size <MB> |
Max repo size limit (default: 200) |
Exit Codes
| Code |
Meaning |
| 0 |
Trust Score ≥ 80 |
| 1 |
Trust Score 40–79 |
| 2 |
Trust Score < 40 |
