Server has 8 tools with consistent naming (verb_noun pattern: list_, get_, analyze_) and reasonable descriptions (100-200 chars). All tools are READ_ONLY, reducing risk. However, input schemas are incomplete: most parameters lack explicit type definitions in the visible code. Parameter descriptions exist but are generic. Output schemas are not documented. Error handling and recovery guidance are absent. Tool composition is sound (each does one thing), but missing pagination details and field-level documentation.

• Document output schemas for all tools. Specify fields returned (e.g., execution_id, workflow_id, status, duration_ms, error_message) with types and whether they are required for downstream calls.
• Add enum constraints to credential_type and status parameters. Replace free-form strings with explicit allowed values (e.g., status: ['success', 'error', 'running', 'waiting']).
• Enhance error descriptions. For each tool, document expected failure modes and recovery steps: 'If workflow_id not found, call list_workflows() first.' or 'If execution still running, retry in 5 seconds.'
• Clarify pagination. Specify whether list_executions returns a total_count, next_cursor, or has_more flag. Document the maximum limit and default behavior.
• Add parameter constraints. Specify min/max for limit (e.g., 1 - 100), format for workflow_id (UUID or alphanumeric), and whether last_n must be positive.

Inferred effective spec: 2026-07-28+.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry