VS Code extension

Six ComfyUI profiles without leaving your editor.

Image, video, audio, 3D meshes, and image understanding — driven from the Command Palette. Every shipped workflow is verified against the live ComfyUI catalog, and missing nodes or models are named before anything is submitted.

Get started Read the Handbook

Command

Ctrl+Shift+P → CodeComfy: Run… (all profiles)

Pick

Profile → preset → the inputs it actually needs

Preflight

Missing model? Named before a run is spent.

Core Capabilities

Six profiles, 27 verified workflows.

Six Profiles

Image, Video, Audio, 3D, Inference, and Metadata — from one command. Pick a profile, pick a preset, answer only the inputs that preset actually needs.

Verified Workflows

The 27 reference graphs are vendored from comfy-headless, where every node type is checked against the live ComfyUI catalog. We do not hand-author graphs — a wrong graph runs green and returns nothing.

Preflight

Nodes checked against /object_info, models against /models. A missing node names its pack, a missing model names its folder — and nothing is submitted, so no GPU time is wasted.

Live Progress

Real sampler steps in the status bar over ComfyUI’s WebSocket — not a static “generating…” for the whole run. Falls back to polling automatically; outputs are never affected.

FFmpeg Optional

Presets ending in CreateVideo → SaveVideo are encoded by the server itself. FFmpeg is only needed for frame-assembly presets.

Workspace Workflow

Outputs save to .codecomfy/ in your workspace — images, video, audio, meshes, and captions alike. Run metadata and index live there for reproducibility.

PNG Provenance

Read the workflow ComfyUI embedded in any output PNG, entirely offline. The fastest way to recover exactly what produced an image somebody sent you.

The Six Profiles

Inputs are derived from each preset’s own graph, not from its profile.

Profile

What it does

Presets

Image

Text-to-image, image edit, union ControlNet

Qwen txt2img, Qwen edit, ControlNet (Qwen / SDXL)

Video

Text- and image-to-video on real temporal models

Hunyuan 1.5 i2v + 720p, Wan 14B, LTX, Mochi

Audio

Text-to-music and stem separation

ACE-Step 1.5 (music / jingle / draft / mp3), separation

Image-to-mesh, exported as GLB

Hunyuan3D-2 (draft / standard / detail)

Inference

Caption, tag, detect, segment, OCR

Florence-2 (7 tasks)

Metadata

Read the workflow embedded in a PNG

Local-only — no server needed

Installation & Setup

Install from VSIX

Extensions → ··· → Install from VSIX…

Download latest from Releases

Configure ComfyUI

Settings → Extensions → CodeComfy

comfyuiUrl: http://127.0.0.1:8188
ffmpegPath: (optional — only for
            frame-assembly presets)

Run anything

Ctrl+Shift+P → CodeComfy: Run… (all profiles)

Profile → preset → inputs.
Preflight names anything missing first.

View Outputs

Outputs saved to .codecomfy/outputs/

Images, video, audio, .glb meshes,
and captions all land here.