MCP workspace · developer access

Make your team context executable.

Connect Claude Code, Codex, Cursor, or another MCP-compatible agent to the work your team already built. Lydo keeps the Space context, permissions, and human approvals in the loop.

The useful distinction

Not another tab for the same work.

Lydo gives external agents a governed front door into shared work. The agent can discover approved context and tools, while the Space remains the authority over what it can read or write.

01
Connect once

Bring the agent to the room.

Use the Lydo MCP endpoint from Claude Code, Codex, Cursor, or another compatible client. Your agent can discover Spaces and work from the same context as the people using the app.

  • One HTTPS endpoint
  • Works with Streamable HTTP clients
  • Use a separate token per device or agent
02
Read with context

Search the work without exporting the workspace.

Approved read tools expose the parts of Lydo your account can already access: Spaces, channels, messages, transcripts, notes, knowledge, boards, and board items.

  • Account and Space scoped reads
  • Knowledge and transcript search
  • Deleted and blocked content stays filtered
03
Write with a human

Useful automation still needs a clear handoff.

Write-capable tokens can send messages or create work, but the default path places requests in the Lydo Outbox. A human reviews the preview before the change runs.

  • Read-only by default
  • Approval queue for ordinary writes
  • Audit activity and immediate revocation
Quick start

Connect Claude Code in one command.

Generate a Lydo MCP token first, then add the HTTPS endpoint to your client. Keep the token private and never commit it to a repository.

For Cursor and other clients, use the same endpoint with Streamable HTTP and an Authorization bearer header. For client-specific configuration,see the developer path

claude mcp add lydo --transport http https://app.lydo.chat/mcp \
   --header "Authorization: Bearer lydo_mcp_..."
Questions buyers ask

Clear enough to choose.

What is the Lydo MCP server?

The Lydo MCP server is a Streamable HTTP connection that lets MCP-compatible clients such as Claude Code, Codex, Cursor, and other agents discover and call approved Lydo tools for Spaces, channels, notes, knowledge, transcripts, and boards.

What can an external agent access?

Access is scoped to the account, token permissions, Space membership, and Space policy. Read-only tokens are the default. A Space can restrict member AI tools, and write requests normally wait in the Lydo Outbox for human approval.

How do I authenticate?

Generate a token in Lydo under Settings, Tool Approvals, and AI Tool Access. The token is shown once and should be passed as a bearer token over HTTPS. Revoke it if it is lost or compromised.

Does MCP let an agent bypass Space permissions?

No. MCP calls run as the connected account and remain subject to token scope, Space membership, admin policy, and approval rules. A connection is a governed front door, not a back door.