MCP server that gives Claude Code hands-on control of NVIDIA Kimodo for AI-powered motion generation. Enables text-to-motion synthesis with game animation presets, model management, and export to FBX/glTF.
Kimodo MCP has well-structured tool definitions with clear names and comprehensive descriptions. All 9 tools follow verb_noun naming (list_*, generate, inspect_*, export_*, cancel_*). Descriptions are detailed (100-300 chars typical), explaining WHAT each tool does and WHEN to use it. However, critical gaps exist: (1) No input schemas visible in source for any tool, only parameter names and descriptions are shown; (2) No output schemas documented anywhere; (3) No error handling guidance or recovery paths; (4) No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels (READ_ONLY, WRITE, REVERSIBLE). The 'generate' tool is complex with 11 parameters but lacks validation rules, constraints, and dependency documentation. Missing pagination for list_* tools. Overall: solid naming and descriptions, but schema and error handling are incomplete.
Cancel a running generation job.
Export a Kimodo BVH motion file to FBX or glTF-binary using Blender. Supports unit/axis presets for Unreal, Unity, or raw output.
Generate motion from text prompts using Kimodo. Supports multiple prompts, custom durations, model selection, diffusion steps, classifier-free guidance, constraints, and post-processing. Returns NPZ, BVH, CSV, and optional demo files.
Get detailed status and results of a specific generation job by ID.
Inspect a Kimodo NPZ motion file for quality metrics: foot skate, ground penetration, jitter, root travel, loopability, and suitability for game animation.
List all generation jobs (completed, running, or failed) with their status, timestamps, and output files.
List all available Kimodo motion generation models with their skeleton types, training data, export formats, and notes.
No input schemas visible in source code. Tool definitions show parameter names and descriptions but no JSON Schema with types, constraints, or validation rules. This violates pattern:constrained-input and pattern:tool-description.
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., job_id from generate() to pass to get_job()). Violates pattern:tool and pattern:response-shaper.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | <=2025-11-25 | v2 |
List all built-in game animation presets (idle, walk, run, combat, etc.) with their prompts, durations, and categories.
Probe the Kimodo environment to check Python version, PyTorch availability, CUDA/GPU status, installed Kimodo version, available models, and offload/cache support.
No error handling guidance. Tools marked WRITE (generate, export_motion) and REVERSIBLE (cancel_job) lack recovery paths, retry guidance, or actionable error messages. Violates pattern:recovery-guide and pattern:error-classification.
Missing tool annotations. No readOnlyHint, destructiveHint, or idempotentHint despite clear risk levels (READ_ONLY, WRITE, REVERSIBLE). Agents cannot infer safety properties. Violates current MCP spec (2026-07-28) tool annotation requirement.
list_* tools lack pagination. No limit, offset, or cursor parameters documented. Large result sets will blow context windows. Violates pattern:paginated-result.