Blender integration through the Model Context Protocol
Critical gaps across all dimensions. Tool descriptions are present but generic (avg ~50 chars, well below 194-char baseline). Input schemas visible for most tools but lack depth: parameters have types but many lack descriptions or constraints. No output schemas documented. Error handling absent, no recovery guidance, retryable classification, or actionable error messages. Security concerns: execute_python accepts arbitrary code with minimal validation (safe_mode check exists but incomplete). Tool composition issues: execute_python is overly broad; set_telemetry_consent is a poor fit for an agent tool. No pagination support on search tools despite potentially large result sets. STDIO transport caps protocol readiness at 50.
Capture a screenshot of the Blender viewport
Execute Python code in Blender
Get a snapshot of the current Blender scene state
Import an asset from Poly Haven into Blender
Import a 3D model from Poly Pizza into Blender
Search for assets on Poly Haven
Search for 3D models on Poly Pizza
Set user consent for telemetry data collection
execute_python accepts arbitrary code with minimal validation. Safe mode check exists but is incomplete and can be bypassed. No confirmation step for destructive operations. Violates pattern:secret-injection and pattern:permission-gate.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields. Violates pattern:tool and pattern:response-shaper.
Search tools (search_poly_pizza, search_poly_haven) lack pagination guidance in descriptions and no result-limit enforcement documented. Large result sets risk context window exhaustion. Violates pattern:paginated-result.
Parameter descriptions missing or generic. 'code' param in execute_python has minimal guidance. 'category' and 'licence' in search_poly_pizza lack enum constraints or valid value documentation. Violates pattern:constrained-input.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 52 | <=2025-11-25 | v2 |
set_telemetry_consent is a poor fit for an agent tool, it controls user preferences, not a task. Should be configuration, not a callable tool. Violates pattern:tool (single responsibility).