Figma Weavy

The CLI toolkit for automating Weavy AI workflows from Python

Figma Weavy is a modular Python CLI toolkit that controls Weavy AI through a lightweight Chrome extension bridge. Create nodes directly, configure model-specific settings, connect workflows, upload media, run image generation, and download results—without repeatedly searching menus or opening side panels.

Build Weavy flows like code: fast, repeatable, and ready for AI agents.

Why Figma Weavy?

• Direct node creation — add cached nodes without searching the Weavy UI.
• Model-aware settings — inspect and change settings exposed by each model.
• Flow automation — create, arrange, connect, run, and inspect workflows.
• Media pipeline — upload image, video, or audio files and download results.
• Deterministic routing — target an exact flow with --flow-id.
• AI-friendly registry — search a compact index, then load one lazy schema.
• No CDP port — Chrome does not need --remote-debugging-port=9222.
• Privacy boundaries — cookie access is restricted to Weavy domains.

Quick start

1. Clone the repository

git clone https://github.com/kodelyx/Weavy.git
cd Weavy

2. Load the Chrome bridge

1. Open chrome://extensions.
2. Enable Developer mode.
3. Click Load unpacked and select the extension/ directory.
4. Sign in to Weavy AI.

3. Start the local backend

python3 -m weavy.bridge_server

Keep that terminal running. The extension connects on port 8765; CLI commands connect on port 8766.

4. Test the connection

python3 -m weavy.cli bridge-status
python3 -m weavy.cli inspect

Build your first automated flow

# Add nodes directly—no sidebar search
python3 -m weavy.cli add Prompt
python3 -m weavy.cli add "ChatGPT Images 2.0"

# Keep the graph compact and connect its two nodes
python3 -m weavy.cli arrange
python3 -m weavy.cli connect

# Inspect model-specific options and update one
python3 -m weavy.cli settings
python3 -m weavy.cli set Quality high

Target an existing Weavy flow explicitly:

python3 -m weavy.cli add Prompt --flow-id YOUR_FLOW_ID

Without --flow-id, Figma Weavy reuses the current in-memory flow, then an open flow tab, and creates a new flow only when none exists.

CLI toolkit

Goal	Command
Check bridge health	python3 -m weavy.cli bridge-status
Inspect the canvas	python3 -m weavy.cli inspect
Add a direct node	python3 -m weavy.cli add "NODE_NAME"
Arrange nodes compactly	python3 -m weavy.cli arrange
Connect two nodes	python3 -m weavy.cli connect
Read model settings	python3 -m weavy.cli settings
Change a setting	python3 -m weavy.cli set "LABEL" "VALUE"
Create a Weavy file	python3 -m weavy.cli create-file
Upload media	python3 -m weavy.cli upload-file /absolute/path/to/media
Generate an image	python3 -m weavy.cli generate-image "PROMPT" --output result.png
Refresh direct actions	python3 -m weavy.cli refresh-catalog
Search node capabilities	python3 -m weavy.cli find-node "text to video" --type video
Load one node schema	python3 -m weavy.cli node-schema "Kling 1.6"
Refresh full schemas	python3 -m weavy.cli refresh-node-schemas
Rebuild AI registry	python3 -m weavy.cli build-node-registry

Run python3 -m weavy.cli --help for the complete command interface.

How it works

Python CLI
   │  localhost:8766
   ▼
Bridge server
   │  localhost:8765
   ▼
Chrome MV3 extension
   │  debugger access scoped to the selected Weavy tab
   ▼
Weavy canvas

The Python backend owns the automation logic. The small Manifest V3 extension only transports browser commands and events. There is no Python runtime, WASM, model catalog, or AI bundled into the extension.

When a command arrives without an open Weavy tab, the bridge opens Weavy, waits for the canvas, and attaches automatically. Browser-mutating commands run one at a time to prevent concurrent tab or flow switching.

AI-agent-ready node registry

Figma Weavy avoids putting a multi-megabyte schema dump into an AI prompt:

1. Search data/weavy_node_index.json locally.
2. Return only a few compact candidates to the AI.
3. Select one node ID.
4. Load only its file from data/weavy_nodes/.
5. Validate ports and model settings before creating or connecting the node.

data/weavy_ai_contract.json describes this tool flow. The full data/weavy_node_schemas.json remains a rebuild artifact and should not be sent to the model.

Project structure

Weavy/
├── weavy/       # Modular Python backend and CLI
├── extension/   # Thin Chrome Manifest V3 bridge
├── data/        # Direct actions, compact index, and lazy node schemas
└── README.md

Security and privacy

• Cookies are scoped strictly to weavy.ai.
• Raw CDP cookie commands are blocked.
• Cookie values are hidden from CLI status output.
• Webpage-origin WebSocket clients are rejected.
• Only one local CLI client can mutate the browser at a time.
• No macOS LaunchAgent, login item, or background startup service is installed.
• Local data/accounts.json, logs, caches, and OS metadata are excluded from Git.

Notes

This is an independent automation toolkit and is not an official Weavy AI or Figma product. Weavy's browser UI and internal behavior may change, so refresh the local catalogs when nodes or models are updated.

Support the project

If Figma Weavy saves you time, star the repository, share it with workflow builders, and open an issue with ideas or bug reports.

Built by kodelyx.