EloquentJS MCP has 21 tools with basic naming and descriptions, but significant gaps in schema completeness and parameter documentation. Most tools lack detailed input schemas, parameters are present but lack type constraints, enums, and validation guidance. Descriptions are present but generic (avg ~80 chars, below the 194-char baseline). No output schemas documented. Error handling is minimal. Tool composition is reasonable (single responsibility mostly observed), but parameter descriptions are sparse. The NLP tools (nlp_query, nlp_crud) are particularly risky, they accept free-form natural language with no validation or error recovery guidance.

• Document output schemas for all 21 tools. Specify returned fields (e.g., list_models returns [{name, table, fields, timestamps}]), types, and which IDs are needed for downstream calls (e.g., model_id for describe_model).
• Add enum constraints to parameters with fixed options. E.g., generate_migration type: enum ['create', 'alter', 'blank']; run_seeder seeder: string with description 'Seeder class name or omit to run all'.
• Expand parameter descriptions to 50-100 chars each. For query_model, clarify: 'Query builder chain as string, e.g., "where(\"status\", \"active\").orderBy(\"created_at\")"'. For run_raw_query, warn: 'Raw SQL query. Use with caution, no validation. Supports SELECT, INSERT, UPDATE, DELETE.'
• Add dry-run or confirmation patterns to destructive tools. E.g., run_migrations: add optional 'dry_run' boolean param. run_raw_query: add 'confirm' boolean param with description 'Set to true only after reviewing the query.'
• Redesign nlp_query and nlp_crud with structured fallbacks. Instead of free-form 'description', accept structured params: model (required), operation (query|create|update|delete), filters (optional object), and provide error messages like 'Could not parse intent. Try: query_model with explicit where/orderBy clauses instead.'