Comfy Cloud CLI for agents — designed with AXI. Generate images, video, and audio via 22 partner models. Submit ComfyUI workflows and templates. Track jobs, estimate credits, search models and nodes, and report usage — all through a shell CLI with token-efficient TOON output.
export COMFY_CLOUD_API_KEY="your-api-key"Get your API key at platform.comfy.org/profile/api-keys.
npx skills add intelligentrascal/comfy-cloud-axi --skill comfy-cloud-axi -gnpx -y comfy-cloud-axinpm install -g comfy-cloud-axi
comfy-cloud-axi setup hooks| Command | Description |
|---|---|
comfy-cloud-axi catalog |
Model type / template tag / node category taxonomy |
comfy-cloud-axi guide <model-or-topic> |
Prompting guide for a model family |
comfy-cloud-axi guide partner |
Full playbook: all 22 partner model slugs + routing |
comfy-cloud-axi technique |
List creative technique recipes |
comfy-cloud-axi technique <name> |
Fetch a step-by-step recipe |
comfy-cloud-axi estimate template <name> |
Estimate credits for a template (read-only) |
comfy-cloud-axi estimate workflow <file> |
Estimate credits for a workflow JSON file (read-only) |
| Command | Description |
|---|---|
comfy-cloud-axi generate <model> "<prompt>" |
Show spend request — no credits charged |
comfy-cloud-axi generate <model> "<prompt>" --confirm |
Generate and spend credits |
comfy-cloud-axi generate <model> "<prompt>" --type video --confirm |
Video generation |
comfy-cloud-axi generate <model> "<prompt>" --aspect-ratio 16:9 --confirm |
With aspect ratio |
| Command | Description |
|---|---|
comfy-cloud-axi job status <prompt_id> |
Point-in-time status snapshot |
comfy-cloud-axi job wait <prompt_id> |
Block until ready (preferred over polling) |
comfy-cloud-axi job cancel <prompt_id> |
Cancel a pending or running job |
comfy-cloud-axi job chain <prompt_id> |
Make output available as input for the next workflow |
comfy-cloud-axi job batch <batch_id> |
Check status of all jobs in a batch |
| Command | Description |
|---|---|
comfy-cloud-axi workflow list |
List saved workflows in the workspace |
comfy-cloud-axi workflow submit <file> |
Submit a workflow JSON → returns prompt_id |
comfy-cloud-axi workflow run <filename-or-uuid> |
Run a saved workflow |
comfy-cloud-axi workflow run <filename> --confirm |
Run with credit spend approved |
comfy-cloud-axi workflow output <prompt_id> |
Retrieve output URLs |
| Command | Description |
|---|---|
comfy-cloud-axi batch submit <file> |
Submit a batch JSON → returns batch_id |
comfy-cloud-axi batch wait <batch_id> |
Block until all jobs terminal |
comfy-cloud-axi batch output <batch_id> |
Retrieve all ready outputs |
| Command | Description |
|---|---|
comfy-cloud-axi templates <query> |
Search templates |
comfy-cloud-axi templates run <name> |
Run template → returns prompt_id |
comfy-cloud-axi templates run <name> --confirm |
Run with credit spend approved |
| Command | Description |
|---|---|
comfy-cloud-axi nodes <query> |
Search ComfyUI nodes |
comfy-cloud-axi nodes <query> --api-only |
Partner / API-backed nodes only |
comfy-cloud-axi models <query> |
Search ComfyUI model library (checkpoints, LoRAs, etc.) |
comfy-cloud-axi upload <file> |
Upload an image |
comfy-cloud-axi queue |
Show account-wide queue status |
comfy-cloud-axi usage |
Workspace spend report |
comfy-cloud-axi usage --group-by product --months 3 |
Spend grouped by product, last 3 months |
22 partner models are reachable via generate. Run comfy-cloud-axi guide partner for the authoritative live list with per-provider routing notes.
Image
| Slug | Description |
|---|---|
bfl/flux-2-pro |
Text-to-image, up to 9 reference images |
bfl/flux-pro-1.1-ultra |
Text-to-image, aspect ratios 1:4 to 4:1 |
bfl/flux-kontext-pro |
Image editing from a prompt |
openai/images-generations |
OpenAI gen/edit — variant via params.model |
vertexai/nano-banana-2 |
Google Gemini 3.1 Flash Image |
vertexai/nano-banana-pro |
Google Gemini 3 Pro Image |
xai/grok-image-generate |
xAI Grok Imagine |
ideogram/generate |
Ideogram text-to-image or edit |
byteplus/images-generations |
ByteDance Seedream — variant via params.model |
Video
| Slug | Description |
|---|---|
bfl/flux-3-video |
BFL FLUX 3 Video + audio |
byteplus/seedance-2.0-t2v |
ByteDance Seedance — tier via params.model |
kling/kling-v3-t2v |
Kling V3 |
minimax/hailuo-03-t2v |
MiniMax Hailuo 03 (H3) |
veo/veo-3-t2v |
Google Veo 3 — tier via params.model |
Audio
| Slug | Description |
|---|---|
elevenlabs/sound-generation |
ElevenLabs sound effect from text |
SVG, 3D, and music use
templates runinstead (e.g.api_recraft_v4_text_to_vector,api_hunyuan3d_text_to_model).
Quick generation (partner model)
guide partner → pick slug
estimate template <name> → confirm cost (optional)
generate <slug> "<prompt>" → spend request (no charge)
generate <slug> "<prompt>" --confirm → prompt_id
job wait <prompt_id> → workflow output <prompt_id>
ComfyUI workflow
workflow list → workflow run <name> [--confirm]
job wait <prompt_id> → workflow output <prompt_id>
Template
templates <query> → templates run <name> [--confirm]
job wait <prompt_id> → workflow output <prompt_id>
Batch
batch submit <file> → batch_id
batch wait <batch_id> (repeat if timed_out: true)
batch output <batch_id>
Chain workflows
workflow run first.json → job wait → job chain <prompt_id>
workflow submit next.json → job wait → workflow output
All commands that spend credits are spend-gated: run without
--confirmfirst to see the cost, then re-run with--confirmafter the user agrees.
pnpm install
pnpm run build
pnpm test
pnpm run devMIT