Integrations

Clave integrates with external services to bring your code and tooling into the same workspace as your issues, docs, and AI assistant. All integrations are configured per workspace and require admin access.

GitHub integration

Connecting a GitHub repository lets Clave index your code, sync issues between systems, and surface relevant context in the AI assistant.

Connecting a repository

Open Settings → Integrations (or navigate to the GitHub section of your workspace settings).
Click Connect GitHub.
Authorize the Clave GitHub app for your organization or personal account.
Select the repository you want to connect.

After authorization, Clave begins indexing the repository. The initial index may take a few minutes depending on the repository size.

What syncs

Once connected:

Repository indexing — file contents and commit history are indexed and made searchable by the AI assistant.
Issue sync — GitHub issues can be viewed alongside Clave issues, keeping your team in one place.
Webhooks — Clave receives push events, pull request events, and issue events from GitHub to keep its index current.

Access token setup

The GitHub integration uses OAuth to authenticate. If your organization uses private repositories, ensure you grant the Clave app access to those repositories during the authorization flow.

If you need to reconnect or reauthorize (for example, after access is revoked), click Reconnect GitHub and go through the authorization flow again.

Google Chat integration

Connecting Google Chat lets Clave send interactive notification cards, respond to AI queries via @Clave mentions, and handle issue actions directly from Google Chat spaces and DMs.

For full setup instructions, feature details, and policy configuration, see the dedicated Google Chat page.

MCP servers

Model Context Protocol (MCP) is an open standard for connecting AI assistants to external tools and data sources. In Clave, MCP servers extend what the AI assistant can do in chat — letting it query external APIs, read data from third-party services, or run custom tools.

The built-in Excalidraw server

Clave includes a built-in Excalidraw MCP server that enables the AI to create and modify whiteboard diagrams. This server is always active and cannot be removed or disabled.

Adding an MCP server

Adding an MCP server requires admin access.

Open Settings → MCP Servers.
Click Add Server.
Fill in the server details:
- Name — a display name for the server (for example, "Sentry" or "Jira").
- Server URL — the URL of the MCP endpoint.
- Transport — choose SSE for legacy server-sent events endpoints, or HTTP for streamable HTTP MCP endpoints. Use HTTP for modern MCP servers; use SSE for older ones.
- Authentication — choose the auth method:
  - No authentication — the server accepts unauthenticated requests.
  - API key — enter your API key. It is stored securely and never exposed in the UI.
  - OAuth / configure URL — for OAuth-based servers, provide an optional access token and configure URL.
- Description — an optional note about what this server does.
Click Add Server.

The server appears in your list immediately. A green dot indicates it is active.

Testing a connection

After adding a server, click Test to verify the connection. The test result shows:

How many tools the server exposes if the connection succeeds
The tool names available
An error message if the connection failed

If the server requires additional configuration (for example, an OAuth flow), a Configure button appears that links to the server's setup page.