MODEL CONTEXT PROTOCOL
Scriben records and understands your meetings. This connects that memory to the assistant you already use, so it answers from what was actually said — not from what you remembered to paste in.
$ npx @scriben/mcp install
NODE 20+ · MACOS · LINUX · WINDOWS
[ Setup ]
No API keys to generate, no config files to hand-edit. The installer writes them for every assistant it finds.
Prints a short code and opens your browser.
Sign in, then type the code from your terminal. The code is what proves it was you who asked.
Claude Desktop and Code, Cursor, Windsurf, VS Code and Zed are written for you. Codex reads TOML, so that one is printed to paste.
Connecting this machine to Scriben… Your code: EV4-KC4 A browser window will open. Sign in and enter that code to approve this device. Connected. Credential stored at ~/.scriben/mcp.json (owner-only). Configured: Claude Desktop — quit and reopen Claude Code — start a new session
[ Tools ]
Your assistant picks whichever it needs. Every response carries the note and date it came from.
list_notesRecent meetings, newest firstget_summaryWhat a meeting covered and decidedget_action_itemsCommitments, and who owns themget_contextWhat you know about a personrecall_memoryFacts carried across meetingssearch_transcriptFind where something was said[ Boundaries ]
A connection to your meeting history is only worth having if its limits are exact.
Nothing can be created, edited or deleted through this connection. There is no write path in the server at all.
Recordings never leave Scriben. Your assistant sees what a meeting meant, not the file.
Every answer names the note and date behind it, so a claim about a person is one you can check.
Remove a device from Integrations in the Scriben app. Access ends on the next request.