MDAN MCP server has significant definition quality gaps. Tool names lack consistent verb prefixes (mdan_list-agents is good, but mdan_agent_* and mdan_ecosystem_* lack clear action verbs). Most tools have descriptions, but many are vague or lack actionable context. Parameter schemas are present but sparse, most tools have 1-3 params with minimal constraint documentation. No output schemas are documented. Error handling is absent from visible code. The dynamic agent tool registration (mdan_agent_*) is inferred rather than explicitly visible, capping those tools at 50. Overall, this reads as a prototype-stage server lacking production-grade polish.

• Add explicit output schemas to every tool. Document the structure of returned content (e.g., 'Returns: {agents: [{name, displayName, title, role}], total: number}').
• Convert parameter descriptions to constraint-based format: 'query (string, 1-100 chars, lowercase alphanumeric + hyphens), keyword to search for'.
• Standardize tool naming to verb_noun pattern: rename 'mdan_ecosystem_catalog' → 'mdan_list_ecosystem_catalog', 'mdan_graph_impact' → 'mdan_analyze_graph_impact'.
• Add error recovery guidance: when a search returns no results, suggest alternative keywords or list available categories. When a file is not found, return 'File not found at {path}. Run ecosystem module installation first.'
• Make dynamic tool registration explicit: generate a static tool registry at startup and register each tool with full schema, or document the schema template for mdan_agent_* and mdan_workflow_* tools.