MCP server for ComfyUI that provides tools for workflow management, asset generation, job tracking, and publishing of generated assets
Server has 15 tools with reasonable naming (verb_noun pattern) and basic descriptions. However, descriptions are often generic and lack actionable context for LLM selection. Parameter schemas are present but inconsistently detailed. Many parameters lack constraints (enums, ranges). Output schemas are not documented. Error handling is minimal, no recovery guidance or categorization. Security considerations (secrets, permissions) are not evident in tool definitions. Composition is reasonable but some tools could be better integrated (e.g., regenerate_image requires asset_id lookup first).
Generate an image using ComfyUI with specified parameters
Get asset record by asset_id with inline preview
Get list of available checkpoint models from ComfyUI
Get current default values for image, audio, and video generation
Get status and results of a ComfyUI job by prompt_id
Get current publish configuration including project root and publish directory
List recent generated assets with optional filtering
List recent jobs with their status
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields (e.g., what does generate_image return? asset_id? prompt_id? Both?). Breaks tool chaining.
Parameter descriptions lack constraints. 'steps' has no range (1-100? 1-1000?). 'cfg' has no bounds. 'width'/'height' have no min/max. LLMs will pass invalid values; no validation guidance in descriptions.
No error handling or recovery guidance. Tools lack descriptions of failure modes, retryability, or next steps. E.g., 'load_workflow', what if file not found? Should LLM call list_workflows first?
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
List available workflow files in the workflows directory
Load a workflow from file and return its JSON structure
Publish a generated asset to the web project directory
Regenerate an image using the same workflow and parameters as a previous generation
Save a workflow to file
Set default values for image, audio, or video generation parameters
Set publish configuration (project root, publish directory, ComfyUI output root)
Descriptions are generic and lack LLM-selection context. 'Get list of available checkpoint models' does not explain WHEN to call this vs other tools, or what the LLM should do with the result. Baseline: 50-200 chars with actionable context.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). LLMs cannot distinguish safe reads from destructive writes without explicit hints. set_defaults, save_workflow, publish_asset should be marked destructive.