What is Supercut for Agents?

Supercut for Agents is a Model Context Protocol (MCP) that gives AI assistants and coding agents permission-aware access to video recordings. It enables semantic search across transcripts, frame-level image retrieval, comment and reaction extraction, and more—all while respecting user permissions. By connecting through a simple terminal command, developers can equip their AI tools with the visual and conversational context stored in recordings, turning passive video archives into active knowledge sources.

Who it's for

• Developers building AI-powered tools – who need to feed their agents with rich, permission-controlled context from recorded meetings, demos, or presentations.
• Product and engineering teams using coding assistants – who want their AI to search, summarize, or reference specific moments in video recordings without manual transcription.
• Knowledge workers relying on recorded content – who need their AI to pull transcripts, frames, or comments from recordings to move work forward faster.

Key features

Permission-aware access

Every request to the Supercut MCP is scoped by user permissions. Agents can only access recordings, transcripts, frames, and comments that the authenticated user is allowed to see. This ensures sensitive content stays protected while still being usable by AI.

Semantic search across recordings

Agents can search recordings by meaning, not just keywords. The MCP understands natural language queries, so you can ask for "the part where the team discussed the marketing budget" and get precise timestamps and context.

Frame and transcript extraction

Pull individual frames as images or full transcripts with speaker timestamps. This lets agents analyze visual content (like slides or whiteboards) alongside spoken words, making recordings fully machine-readable.

Comment and reaction retrieval

Agents can list comments and reactions left on recordings. This turns asynchronous feedback into actionable data—your AI can summarize team sentiment, flag unresolved discussions, or highlight popular moments.

What stands out

"Supercut gives your agents eyes and ears—permission-aware access to recordings so they can move work forward, not just read text."

The core differentiator is that Supercut treats recordings as first-class data sources for AI, not just playback files. While most MCPs focus on text or code, Supercut bridges the gap between visual/audio content and agent reasoning. The permission-aware design means teams can safely integrate recordings into their AI workflows without compromising privacy or access controls.

Worth checking out if…

You're building or using AI assistants that need to understand recorded meetings, presentations, or demos—especially if you want to search, summarize, or extract specific moments from video content while keeping permissions intact.