Diagnoses failed ComfyUI workflow executions: extracts errors from history/logs, inspects nodes/models, identifies causes, suggests fixes. Use prompt_id or 'last'.
From comfynpx claudepluginhub artokun/comfyui-mcp --plugin comfyprompt_id or "last" to debug the most recent execution/debugDiagnoses Claude Office add-in deployment issues like stale configs, connection failures, missing add-ins, and auth errors via admin triage playbook.
/debugRuns autonomous bug-hunting loop with scientific method and autoresearch iterations to find all bugs. Supports --fix, --scope, --symptom, --severity, --iterations flags.
/debugInvokes oac:debugger skill to diagnose bugs, test failures, or unexpected behavior exactly as specified, before any fix proposals.
/debugOrchestrates four specialist agents to systematically debug a code task description, producing problem analysis, refined hypotheses, diagnostic plan, and user confirmation request.
/debugPerforms systematic debugging on described code issues via 6-step process: reproduce, isolate, analyze, hypothesize, test, fix. Outputs debugging plan, root cause, fixes, prevention, and test cases.
/debugInvokes debug-like-expert skill to systematically debug the specified issue via evidence gathering, hypothesis testing, and verification.
The user wants to find out why a ComfyUI workflow execution failed and get a suggested fix.
Get the execution history. The argument is: $ARGUMENTS
get_history with that prompt_idget_history with no prompt_id to get the most recent executionExtract the error. From the history response, identify:
node_id: the node where execution failednode_type / class_type: what kind of node it wasexception_message: the error messagetraceback: the full Python tracebackIf the execution succeeded (no error), tell the user it completed normally and show the output summary.
Get relevant logs. Call get_logs with a keyword filter matching the error — use the exception class name (e.g., "RuntimeError", "ValueError") or a distinctive phrase from the error message. Request up to 200 lines.
Inspect the failing node. Call get_node_info with the node_type of the failing node. Check:
Check for missing models. If the error mentions a missing file, model, or checkpoint:
list_local_models to see what's installeddownload_model or provide a search query for search_modelsCheck for missing nodes. If the error is a KeyError or mentions an unknown node type:
search_custom_nodes with the node class name to find which pack provides itPresent the diagnosis. Provide a clear summary:
clear_vram before retryingUser: /comfy-debug last
Steps:
get_history with no prompt_idget_logs with keyword "RuntimeError"get_node_info for "KSampler"clear_vram before retrying