Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/artokun/comfyui-mcp/debuggit clone --depth 1 https://github.com/artokun/comfyui-mcpWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00009 | $0.00920 |
| Opus 5 | $0.00005 | $0.00460 |
| Sonnet 5 | $0.00002 | $0.00184 |
| Haiku 4.5 | $0.00001 | $0.00092 |
Grade A, and why
debug scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
Copies of this mod
1 near-identical copy found in the catalogue:
- debug — 86% identical, 32 lines differ
How it starts
The opening of the file, as written. The whole thing — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/comfy-debug — Diagnose a Failed Workflow
The user wants to find out why a ComfyUI workflow execution failed and get a suggested fix.
Instructions
-
Get the execution history. The argument is: $ARGUMENTS
- If a prompt_id is provided, call
get_history(action="list")with that prompt_id - If "last" is provided or no argument given, call
get_history(action="list")with no prompt_id to get the most recent execution
- If a prompt_id is provided, call
-
Extract 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 traceback
If the execution succeeded (no error), tell the user it completed normally and show the output summary.
-
Get relevant logs. Call
get_system_stats (action:"logs")with akeywordfilter 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
create_workflow (action:"node_info")with thenode_typeof the failing node. Check:- What inputs does the node expect?
- Are there required inputs that might be missing or mistyped?
- What data types are expected for each input?
-
Check for missing models. If the error mentions a missing file, model, or checkpoint:
- Call
list_local_modelsto see what's installed - Compare against what the workflow references
- If a model is missing, suggest using
download_modelaction:"download"or provide a search query foraction:"search"
- Call
-
Check for missing nodes. If the error is a
KeyErroror mentions an unknown node type:- The node pack may not be installed
- Call
search_custom_nodes(action: "search") with the node class name asqueryto find which pack provides it - Suggest installing the pack
-
Present the diagnosis. Provide a summary:
- What failed: node type, node ID, and a one-line description
- Why it failed: root cause explanation in plain language
- Suggested fix: concrete steps the user can take to resolve the issue
- Traceback excerpt: the most relevant lines from the traceback (not the full dump)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 73 lines · 9 tokens per session scan A 31f82a9175bd
debug is a command published in the GitHub repository artokun/comfyui-mcp (701 stars, last pushed 2d ago), licensed MIT. It adds 9 tokens to every session and 920 once invoked, about $0.0000 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
setup
You are setting up the Video Recreation Agent Team inside this comfyui-agent repo. Run every step below automatically. Don't ask permission — just do it, report what happened.
PRODUCTION_HARDEN
Each agent is a Claude Code sub-agent invoked via claude --model claude-sonnet-4-6-20250929 with a role-specific prompt file. Agents operate on branches, submit PRs via conventional commits.
PRODUCER
You are the production manager for this open-source VFX tool. You handle everything that's NOT the code itself: CI/CD, packaging, release management, documentation infrastructure, metrics, and cross-agent coordination. You're the one who makes sure the train runs on time and nothing ships broken.
qa-compare
You are the QA expert in the Video Recreation Agent Team.
COMFY_LEAD
You are the ComfyUI technical lead. You know ComfyUI's API, node system, workflow JSON format, and the MCP protocol deeply. You're hardening an AI co-pilot that helps VFX artists work with ComfyUI through natural language.
NUKE_COMP
You are a senior VFX compositor with deep Nuke pipeline experience who has transitioned into pipeline TD work. You think about tool integration the way a compositor thinks about a comp tree — everything must connect cleanly, data must flow predictably, and the artist should never have to think about the plumbing.