Critical gaps across all dimensions. No visible input schemas, parameter descriptions, or output documentation for any of the 6 tools. Tool definitions appear to be inferred from hook file paths and brief descriptions rather than explicit MCP registrations. Descriptions are present but minimal (10-100 chars), lacking WHAT/WHEN/WHY context. No evidence of error handling guidance, parameter validation, or response shaping. This server exposes governance state mutations (onboard, sync_state, bind_session) without documented schemas or safety constraints.

• Add explicit JSON Schema input definitions for all 6 tools. Include type, description, and constraints for every parameter (e.g., uuid format, token length, session_id pattern).
• Expand tool descriptions to 50-200 chars. Answer: What does it do? When should the LLM call it instead of a similar tool? What does it return? E.g., 'Synchronize agent state with governance server. Call after agent completes a task to reset milestone accumulator and update last_checkin_ts. Returns updated session binding.'
• Document output schemas for onboard, start_session, bind_session. Specify which fields (UUID, token, session_id, label) are returned and their types so downstream tools can chain calls.
• Add parameter descriptions for all inputs. E.g., if onboard accepts a 'label' param, describe: 'Human-readable label for this governance identity (1-255 chars, alphanumeric + underscore).'
• Implement error handling guidance. For onboard failures, return: 'Onboarding failed: [reason]. Retry with a different label or contact governance admin.' For sync_state, clarify which state mutations are retryable vs fatal.