Stirrup provides 7 tools with clear verb-noun naming (list_, get_, create_, run_, validate_) and reasonable descriptions (avg ~120 chars). All tools have input schemas with type declarations. However, parameter descriptions are sparse or missing entirely, most params lack guidance on format, constraints, or valid values. Output schemas are undocumented. Error handling is minimal; no recovery guidance or actionable error messages visible. The sampling feature (deprecated) is enabled but not critical. Overall: solid foundation, but parameter and output documentation gaps prevent higher scores.

• Add parameter descriptions to all tools. For 'workflow' params, specify: 'Complete workflow definition as a JSON string. Must include: id (string), name (string), version (string), nodes (array of node objects with id, type, config), edges (array of edge objects with source, target).'
• Document output schemas for all tools. E.g., list_workflows returns: {workflows: [{id, name, description, parameterInfo}], total: number}. get_workflow returns: {id, name, version, nodes, edges, config}.
• Add error handling guidance to descriptions. E.g., create_workflow: 'Returns validation errors if workflow is malformed. Common errors: missing required fields, invalid node types, circular edges. Call validate_workflow first to debug.'
• Mark destructive tools with explicit state-change language. E.g., run_workflow: 'Executes and modifies workflow state. Returns execution ID and initial state. Use get_execution to poll for completion.'
• Add pagination to list_workflows and list_node_types. Include limit (default 20, max 100) and offset parameters. Return total count and next_offset in response.

Inferred effective spec: <=2025-11-25.

• Relies on Sampling (deprecated) - integrate directly with the LLM provider API