Use SquadBear from Claude — the MCP integration

Connect Claude to SquadBear over MCP and run leave, time, team health and core HR in plain language — with your own permissions.

SquadBear ships a built-in MCP server, so Claude can read and act on your workspace in plain language — using your own login and your own permissions. Nothing new to install; you add SquadBear as a custom connector and start asking.

What you can do with SquadBear in Claude

Anything you could do in the web app, you can ask Claude to do instead. The connector exposes over 100 tools across the whole product: booking and checking leave, logging time and pulling time reports, running team-health checks, managing employee records, onboarding after you've hired someone, and chasing policy sign-offs. Because it is the same permission model as the UI, the answer is always grounded in your real data — Claude runs the calculation, it doesn't guess.

What it does not do: emotion or mood inference, candidate screening, ranking people for hire/fire/promote, or automatic people decisions off a score. Claude is the assistant; SquadBear is the system of record and the permission boundary.

A useful first move is to ask Claude to call whoami. It should come back with your role, manager, teams and holiday calendar — proof the connection resolved to you before you hand it anything real.

How to connect Claude to SquadBear

Setup takes a minute and lives entirely in Claude's settings:

1. Your MCP endpoint is https://app.squadbear.com/mcp — it's also shown on the in-app Agents page for your own workspace.
2. In claude.ai or the Claude desktop app, open Settings → Connectors → Add custom connector (custom connectors need a paid Claude plan). Name it "SquadBear" and paste the endpoint.
3. Claude opens the SquadBear sign-in. Log in, then choose which scopes the connection may use — things like leave:read, time:write or health:write.
4. In any chat, enable SquadBear under the search & tools menu, and start asking.

The full walkthrough — with the scope list and a worked example — is in Connect Claude to SquadBear.

Permissions, scopes and the audit trail

The agent acts as you — with your role's permissions, nothing more. The scopes you approve are frozen into that connection; to widen them later you disconnect and reconnect. Every action Claude takes is written to the same audit log as the web app and the REST API, so nothing it does is invisible to an admin looking at Admin → Audit. This is the design rule behind SquadBear: the UI, the API and the assistant all go through one permission and audit path, so an agent is never a back door.

Prompts to try

Once connected, these are the kinds of things you can ask straight away:

Related

• Connect Claude to SquadBear — the full setup guide
• Prompts to run through your connected agent
• Use SquadBear from ChatGPT