comfyui-weaver
π§΅ Weaves Claude Code into ComfyUI - an MCP bridge that runs your renders (local + Comfy Cloud) without disturbing your work, QCs them by actually looking at them, and ships a script-to-screen film pipeline on top.
Install / Use
claude mcp add GuusF -- npx -y github:GuusF/comfyui-weaverIf the server publishes to npm under a different name, use that package instead β check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
AutomationSupported Platforms
Tags
Our assessment of comfyui-weaver
comfyui-weaver scores 72/100 on our quality scale, 2638th of 2,891 Automation skills we index.
Its MCP Server is 5.6 KB long, well organised into 10 sections with 3 code examples: a solid amount of guidance for an agent.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated about 4 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 90/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit β read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-08. Automated pattern scan on 2026-10-08. It catches known dangerous patterns, not every risk β read a skill before letting an agent act on it.
comfyui-weaver compared with similar skills
All 4 of these similar skills score higher than comfyui-weaver; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| comfyui-weaver (this skill)by GuusF | 72 | 3 | 4mo ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 93.2k | today | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 74.6k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.3k | today | CLAUDE.md |
| Scraplingby D4Vinci | 100 | 86.2k | today | MCP Server |
Frequently asked questions
- How do I install comfyui-weaver?
- Run
claude mcp add GuusF -- npx -y github:GuusF/comfyui-weaver. The install tabs above show the steps for each supported agent. - Which AI agents does comfyui-weaver work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is comfyui-weaver safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 90/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is comfyui-weaver still maintained?
- The repository was last updated about 4 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.
Skill content
View source on GitHubcomfyui-weaver π§΅
Weaves Claude Code into your ComfyUI β local and Comfy Cloud β running your renders quietly in the background without ever tangling your threads.
A self-contained MCP server that lets Claude Code operate a ComfyUI installation: discover nodes and models, build/convert/run workflows, track jobs, look at the results (Claude reads the images it generates), run parameterized templates on Comfy Cloud, and orchestrate a full scriptβstoryboardβvideo film pipeline on top.
Claude Code ββ(stdio MCP)ββ comfy_mcp_server.py βββ¬β(HTTP/WS)ββ local ComfyUI
ββ(HTTPS)ββββ cloud.comfy.org
The non-disruption contract
Built for setups where ComfyUI is a working tool, not a toy. The server:
- never starts, stops, or restarts ComfyUI β it attaches to what's running;
- always queues at the back (never
front), never clears queue/history; - namespaces everything it generates under
output/claude/andinput/claude/β refuses workflows whose save paths it can't contain; - only cancels jobs it submitted itself (unless you
force); - never frees VRAM or unloads models behind your back;
- submits cloud jobs (your credits!) only on explicit request.
Install (Windows, PowerShell)
-
Clone β inside your ComfyUI data directory is simplest:
cd <your ComfyUI data dir> # the folder holding models/, output/, user/ git clone https://github.com/GuusF/comfyui-weaver claude-integration powershell -ExecutionPolicy Bypass -File claude-integration\install.ps1The installer creates a private venv (your ComfyUI's Python is never touched), installs deps (
mcp,httpx,websockets,pillow), writes a.mcp.jsonfor Claude Code, and runs an offline self-test. -
Cloned somewhere else instead? Set the data dir explicitly in
.mcp.json:"env": { "COMFY_DATA_DIR": "D:\\path\\to\\ComfyUI" } -
Restart Claude Code in that folder and approve the
comfyuiserver. Say "check comfy status" to verify.
Ports: auto-detects 127.0.0.1:8000 (Desktop) then :8188 (standalone);
override with COMFYUI_URL.
Comfy Cloud (optional)
Create an API key at platform.comfy.org (paid tier required for API access;
the full key is shown only once) and save it as the single line of
state/cloud_api_key.txt (git-ignored, picked up without restart) β or set
COMFY_CLOUD_API_KEY. Every cloud run consumes your credits; the bridge
makes Claude state costs before submitting.
What Claude can do with it
22+ tools, the highlights:
| | |
|---|---|
| comfy_status, search_nodes, node_info, list_models | discover what's installed |
| run_workflow (API or UI-format JSON, or a named template) | execute, with validation + output containment |
| wait_for_job, job_status, cancel_job, queue_info | job lifecycle, queue-aware |
| view_output | Claude sees the result inline and can QC it |
| history_to_template | turn any past run into a reusable parameterized template β the fastest path to a working pipeline |
| convert_ui_to_api | editor JSON β API JSON (handles dict-form widgets_values, reroutes, primitives) |
| get_logs | server log tail for debugging failed runs |
| cloud_* (status, models, run, wait, upload, cancel) | the same flows on Comfy Cloud GPUs |
Templates ship in templates/ (text-to-image, image-to-video via Kling
partner nodes, tiled Flux 2 upscaling with LoRA slots) and are plain JSON β
add your own or let Claude capture them from history.
The film pipeline (optional, fun)
film/ + docs/skills/film-production/ implement a gate-based short-film
pipeline: script β shot manifest (single source of truth) β look dev β
storyboard keyframes β animatic ("lock pacing before money") β per-shot
video takes (local or cloud) β vision-QC dailies (Claude reads frames and
flags retakes) β assembly + CMX3600 EDL for your NLE. See
film/PIPELINE.md.
Claude Code knowledge files
docs/skills/ contains skill files that teach Claude the workflow JSON
format, tool recipes, and pipeline etiquette. Copy them into your project:
Copy-Item -Recurse docs\skills\* <data dir>\.claude\skills\
docs/CLAUDE.example.md is a starting point for the data dir's CLAUDE.md
(the rules Claude follows there) β edit it to match your machine.
Verify
.venv\Scripts\python.exe tests\smoke_test.py # full (needs ComfyUI running)
.venv\Scripts\python.exe tests\smoke_test.py --offline # protocol + imports only
.venv\Scripts\python.exe tests\cloud_test.py # cloud auth + catalog (free)
The live test queues a LoadImageβSaveImage passthrough β no GPU, no models,
and verifies output lands in output/claude/ with your queue untouched.
Security notes
state/(API key, job tracking, logs) is git-ignored β keep it that way.- The API key file is plain text on your disk; rotate it at platform.comfy.org if it ever leaks. Never commit it.
- The MCP server binds nothing: it's stdio-only, spawned by Claude Code.
License
MIT β see LICENSE.
Related Skills
Agent-Reach
93.2kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu β one CLI, zero API fees.
headroom
74.6kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
47.3kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Scrapling
86.2kπ·οΈ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit β see the Safety scan above for what the skill file itself contains.
