Blender MCP has 16 tools with basic naming and descriptions, but critical gaps in schema completeness, parameter documentation, and error handling. Most tools lack detailed input parameter descriptions and output schema documentation. The execute_code tool is particularly risky with minimal safeguards. Descriptions average ~60 chars (well below the 194-char baseline for A+ tools). No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels. Parameter descriptions are sparse or missing entirely.

• Add tool annotations to all tools: mark read-only tools with readOnlyHint=true, destructive tools (execute_code, downloads, imports) with destructiveHint=true, and idempotent tools with idempotentHint=true.
• Expand tool descriptions to 100-200 chars. Current descriptions (~50-60 chars) lack context on when to use each tool vs alternatives. E.g., 'Get available PolyHaven asset categories (textures, models, HDRIs, etc.) to guide search queries' instead of 'Get available PolyHaven asset categories'.
• Document output schemas for all tools. Specify what fields each tool returns (e.g., get_scene_info returns {objects: [{name, type, location}], frame_range: {start, end}}). This enables LLMs to chain tools correctly.
• Add parameter constraints and formats. For search_polyhaven_assets, specify: query (1-100 chars, free text), category (enum: textures|models|hdris|brushes). For create_rodin_job, specify prompt length limits and supported asset types.
• Implement confirmation/dry-run for execute_code. Add a 'dry_run' boolean parameter and require explicit confirmation before executing arbitrary Python. Return what would execute without side effects.

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

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