Recordings your AI agent can actually use.

Stop typing descriptions and quickly create transcripts, captioned frames, and artifacts...all AI-agent ready.

LibraryDetail viewMCP integrationExtension
ClipCabinet
⌘K
figma-layout-debug4:18

Walked through the Inspector after the alignment broke on the header. Found the issue at 2:14.

figma.comfigmacss
prisma-migration-fail2:41

Reproduced the migration error against staging; captured the stack and the env var that was missing.

github.compostgresops
claude-mcp-handshake6:02

Recorded the agent's full tool-call sequence end-to-end, from list_recordings to get_recording with citations.

claude.aimcp
MCP integration

Wire it into your agent in one paste.

.mcp.json
{
  "mcpServers": {
    "clipcabinet": {
      "type": "http",
      "url": "https://mcp.clipcabinet.com",
      "headers": {
        "Authorization": "Bearer kg_xxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}
Sample tool call
> what was the error I hit at the end of the figma walkthrough?

↳ At [2:14] you opened the Inspector and saw:
   "Cannot read properties of undefined (reading 'fills')"
   It came after you tried to copy a layer style from a hidden frame.
   Source: recording figma-layout-debug, artifact #3.

Why ClipCabinet

Recordings are first-class context.

Not video files in a folder. Searchable transcripts, captioned frames, extracted artifacts.

Your agent sees what you saw.

Transcripts, screenshots, URLs and errors lifted out for you. Addressable from any MCP client.

Bring your own keys.

Your Anthropic, OpenAI, and Deepgram keys. Your own R2 or S3 bucket. No middleman.