A local AI studio that edits, not just generates

What it does

NodeTool is a desktop app that acts as an end-to-end AI production studio for video, image, audio, and text. You pitch a concept to an inbuilt agent, which drafts a script, generates storyboard stills, renders clips, and assembles them on a multi-track timeline. The output is a living project you can revise frame by frame in its node editor, timeline, sketch pad, or script editor.

The interesting bit

The agent doesn’t live in a chat panel—it drives the node canvas, timeline, sketch editor, and script editor directly through exposed actions. Every surface doubles as an MCP tool, so you can point Claude Desktop or Codex at the local server and let your own agent run the studio. The result is a production pipeline where the agent builds the foundation, but you retain the edit.

Key highlights

• Multi-track video editor with regeneratable AI clips, not a black-box export.
• Storyboard-first workflow: generate cheap stills, pick shots, then render clips; revise one shot without re-rolling the whole reel.
• Script editor with per-character voice casting, take management, and word-level sync sent to the timeline.
• Runs fully offline with local inference via Ollama, MLX, and GGUF, or plugs in cloud provider keys with no markup.
• AGPL-3.0 end-to-end; custom nodes in TypeScript or Python, plus an MCP server to integrate external agents.

Verdict

Worth a look if you want an open, local alternative to black-box generative video tools and don’t mind steering an agent through the creative process. Skip it if you just need a quick one-shot image generator or a simple chat interface—this is a full production suite with a learning curve.