io.github.levelsofself/nervous-system — Model Context Protocol (MCP) Governance Server
This MCP server provides governance for LLM agents via “7 enforced rules” and accompanying read-only reference tools. It is described as deterministic, lint-like, and safe across MCP clients, focused on validating MCP server configurations. It includes tooling to audit configuration risks and references to protected files.
🛠️ Key Features
- Enforced governance rules: 7 rules
- Deterministic governance lint for MCP server configurations
- Read-only reference tools
- Offline operation and deterministic results (no dependencies)
🚀 Use Cases
- Auditing MCP server configurations for security issues
- Checking for inline secrets, insecure transports, and unsafe scopes
- Running preflight checks using a configured protected files list
⚡ Developer Benefits
audit_mcp_config: deterministic lint, fully offline, never throws; returns findings with severities and a 0–100 scorecheck_preflight: readsprotected_files_listfromnervous-systemwhen configured
⚠️ Limitations
check_preflightonly performs work when the server is configured- No behavior beyond reference tooling and configuration lint is described in provided material