Contralabs open-sources 234-step Premiere dataset for training video-editing agents
yuwen_lu_ · x · 2026-07-21
Contralabs has open-sourced a preview dataset for training video-editing agents.
The dataset contains 234 annotated steps across 4 computer-use trajectories, captured from professional editors working in Adobe Premiere Pro. Each session includes more than 2 hours of trajectory data, hundreds of edits in just four sessions, audio of think-aloud reasoning, native tool-use details, and real video assets rather than synthetic examples. The data is formatted for an AgentNet-style trajectory schema with Premiere-specific extensions.
More from coding & agent
- HeyGen adds a media-sourcing skill for coding agents with 75k images and 10k tracks — HeyGen · 2026-07-22
- Agent search bottlenecks are now about variance, not raw latency — rohanpaul_ai · 2026-07-22
- LangSmith adds tracing for Pipecat, LiveKit, OpenAI Realtime, and Gemini Live — LangChain · 2026-07-22
- An MCP server signs every AI agent tool call into a verifiable Merkle chain — Funky_Chicken_22 · 2026-07-22
- Annotated transcript of a Claude Code team interview is now available — trq212 · 2026-07-22
- Claude Code skill uses 10 Markdown rules to make outputs ADHD-friendly — alex_verem · 2026-07-22