Skip to content

yutianxiao6/comfyui-mcp / skills/comfyui-workflow

comfyui-workflow

Verified

Build, inspect, modify, validate, run, and troubleshoot ComfyUI workflows through ComfyUI MCP tools. Use for requests involving ComfyUI nodes, models, workflow JSON, graph connections, queue execution, generated outputs, canvas synchronization, LoRA, ControlNet, inpainting, upscaling, image-to-image, video, audio, 3D, or custom nodes.

references/troubleshooting.md

View on GitHub
# Troubleshooting

Identify the layer first. Avoid rebuilding an entire workflow for a local error.

## MCP server does not start

- Confirm Python 3.11+ and `uv` are available for source/plugin execution.
- Run `uv run comfyui-mcp --help` from the plugin or repository root.
- Check the host's MCP logs for command, working-directory, dependency, or JSON protocol errors.
- Keep stdout reserved for MCP protocol messages; diagnostic output belongs on stderr.

## ComfyUI is unreachable

- Call `comfyui_health` and use its configured base URL in the diagnosis.
- Confirm ComfyUI is running and reachable from the MCP server process.
- If ComfyUI listens on a non-default URL, set `COMFYUI_URL` in the host's MCP configuration and restart the MCP connection.
- Do not infer that a timeout means the workflow is invalid.

## Unknown class type

- Search with `comfyui_list_node_types`.
- Check spelling and case with `comfyui_get_node_type`.
- If absent, report the missing custom-node package or choose only from verified alternatives approved by the user.

## Missing or invalid input

- Inspect the current graph and live node schema.
- Distinguish a literal from a link array.
- Check required versus optional inputs, exact enum values, numeric ranges, and model filenames.
- Revalidate after the smallest correction.

## Link type or output-index failure

- Inspect both source and target schemas.
- Verify the output index is zero-based and exists.
- Verify source output and target input types are compatible.
- Disconnect the incorrect input, reconnect it, and validate.

## Prompt rejected by ComfyUI

- Preserve the returned validation details.
- Check every node named in the error against `object_info` through `comfyui_get_node_type`.
- Confirm files and models exist on the ComfyUI host, not merely on the agent host.
- Do not retry with `validate=false` unless the user is deliberately debugging ComfyUI's raw rejection behavior.

## Run is failed or unknown

- Query `comfyui_get_run` with the exact prompt ID.
- Inspect execution error details and the current queue.
- An `unknown` result can mean history was cleared, the wrong ComfyUI instance is configured, or the prompt ID came from another server.
- Preserve failed-run evidence before submitting a replacement run.

## Canvas does not update

- Confirm `comfyui_health` reports the optional bridge.
- Confirm a ComfyUI browser tab with the bridge is open and connected.
- Retry synchronization only after connectivity is restored; do not recreate the managed workflow.

## Revision conflict

- Read the latest workflow.
- Compare the desired mutation with the current graph.
- Apply only the missing edit using the latest revision.
- Never suppress the guard by replaying a stale batch blindly.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall