Label Studio
ActiveAn AI agent for Label Studio. It works in a scene with a live connection to Label Studio. It asks before it writes, and every run is traced.
Label Studio is an open-source data labeling platform for text, images, audio, and video. Connect it to queue work for human annotators, read back completed labels as ground truth, and close the loop between an agent's output and a human's correction. Works against Label Studio Cloud or a self-hosted instance.
What you can connect
Add these to your scene and AI gets access.
Label Studio instance credentials for data labeling
A Label Studio labeling project
Connect Label Studio over MCP
Daslab is itself an MCP server.
Connect Label Studio in a scene, then pin that scene's URL in Claude Code, Cursor, or any MCP client. The agent there gets your Label Studio tools, and your approval gate and trace travel with them.
https://daslab.run/mcp/<your-world>/<your-scene>One scene can hold Label Studio and everything else you connect, so a single endpoint carries them all. Set up a client.
Use cases
- Queue an agent's uncertain extractions for human review, then read the corrections back as ground truth
- Build an eval set by importing documents, having humans label them, and exporting the annotations to a grader
- Track labeling progress across projects and report when a dataset is ready
- Feed reviewed labels into a benchmark's expected-output fixtures
From Label Studio
Similar integrations
Others in AI & ML.
Ready to try Label Studio with Daslab?
Get started with CLI