Documentation
Integrate your AI tools with Tanjiren.
MCP Server
Tanjiren exposes a remote Model Context Protocol (MCP) server. Any MCP-compatible client can connect to it and interact with your Tanjiren organization — list workstations, dispatch commands, manage prompts, and more — all from your AI assistant.
AI-readable docs
Tanjiren now also exposes public, agent-friendly documentation so MCP clients and AI systems can discover the platform without scraping the visual UI.
Connect
The only thing you need is the server URL:
https://mcp.tanjiren.app/mcpClaude Code
claude mcp add --transport http tanjiren https://mcp.tanjiren.app/mcpCursor
Open Settings → MCP Servers → Add Server. Select "HTTP" transport and paste the URL above.
Other MCP clients
Add the URL as a Streamable HTTP remote server. Tanjiren follows the MCP specification — any compliant client will work.
Your client will open a browser window for you to sign in with your Tanjiren account. After authorization, you're connected — no API keys or manual configuration needed.
Authentication
Tanjiren uses OAuth 2.1 with PKCE. The flow is fully automatic:
- Your MCP client discovers the OAuth server via
.well-known/oauth-protected-resource - It registers itself via Dynamic Client Registration (RFC 7591)
- A browser window opens for you to sign in with your Tanjiren account
- The client receives a scoped access token — no secrets to manage
Available tools
The tools available to you depend on your organization role:
| Tool | Description | Scope |
|---|---|---|
| whoami | Your identity, organization, and role | mcp:read |
| list_workstations | All workstations in your organization | mcp:read |
| get_workstation | Details and status of a specific workstation | mcp:read |
| list_prompts | Saved prompts in your organization | mcp:read |
| list_dispatch_runs | Recent dispatch runs | mcp:read |
| get_dispatch_run | Details of a specific dispatch run | mcp:read |
| list_org_members | Members of your organization | mcp:read |
| list_audit_events | Audit log (admin only) | mcp:governance:read |
| create_dispatch_run | Dispatch a prompt or command to workstations | mcp:operate |
| cancel_dispatch_run | Cancel a running dispatch | mcp:operate |
| create_prompt | Save a new prompt | mcp:operate |
| delete_prompt | Delete a saved prompt | mcp:operate |
Available resources
In addition to tools, Tanjiren exposes read-only MCP resources for stable context that hosts can cache, inspect, and hand to the model without treating every read as an action.
| Resource | Type | Description |
|---|---|---|
| tanjiren://session/whoami | static | Current user, org, plan, and granted scopes. |
| tanjiren://org/current | static | Current organization summary and available capabilities. |
| tanjiren://docs/mcp-overview | static | Canonical MCP usage guidance for an AI client. |
| tanjiren://workstations/{instanceId} | template | Canonical workstation record for one workstation. |
| tanjiren://dispatch/runs/{runId} | template | Canonical dispatch run record for one run. |
Scopes and permissions
Access is granted based on your organization role. Your MCP client may request scopes, but the server clamps them to what your role allows:
| Role | mcp:read | mcp:operate | mcp:governance:read |
|---|---|---|---|
| Viewer | Yes | — | — |
| Member | Yes | Yes | — |
| Admin | Yes | Yes | Yes |
| Owner | Yes | Yes | Yes |
Example usage
Once connected, just ask your AI assistant naturally: