MCP — Connecting AI Clients

MCP (Model Context Protocol) is an open standard that lets AI assistants like Claude Desktop interact directly with your Linc AI data. Through MCP, external clients can query your initiatives, sessions, processes, and knowledge base without leaving their own interface. Access is controlled through personal access tokens you create in Settings.

A note on terminology: the API and its tools and scopes use the word project for the workspace container the app labels an Initiative — they are the same thing. For example, list_projects returns your Initiatives.

Creating a personal access token

1. Go to Settings in the left sidebar, then open the Developer group and click MCP Tokens.
2. Enter a Name for the token (e.g. "Claude Desktop") and set the expiry in days (1–365, default 30).
3. Select the scopes the token needs. For read-only access — the recommended setup for most uses — check only the :read scopes.
4. If you selected any write scopes, an additional confirmation is required: check the acknowledgement box and type ENABLE_WRITE_SCOPES.
5. Click Create Token. The token is displayed once — copy it immediately and store it securely. If you lose it, revoke it and create a new one.

Available scopes

Grant only what your client needs. Read scopes: mcp:projects:read, mcp:collections:read, mcp:files:read, mcp:sessions:read, mcp:processes:read, mcp:sop:read, mcp:bpmn:read, mcp:fitgap:read, mcp:people:read, mcp:plan:read. Write scopes: mcp:collections:write, mcp:files:write, mcp:sessions:write, mcp:plan:write.

Available tools

Once connected, your MCP client can call tools including:

• Initiatives: list_projects, get_project
• Collections & files: list_collections, get_collection, create_collection (write), list_files, get_file, get_file_text, upload_text_document (write)
• Sessions: list_sessions, get_session, get_session_messages, list_session_files, create_session (write), list_session_assignments, create_session_assignments (write)
• Processes: get_process_hierarchy, get_process_hierarchy_node, list_process_gaps, get_process_strategy_draft, get_process_chain_draft, list_process_redesigns
• SOP & BPMN: get_session_sop, export_session_sop, get_session_bpmn, export_session_bpmn
• Fit/Gap: list_fitgap_runs, get_fitgap_run, list_fitgap_results, get_fitgap_result, reconcile_process_with_sop
• People & interview plans: list_org_people, get_org_person, get_org_understanding, list_interview_plans, get_interview_plan, send_interview_plan (write)
• Knowledge base: search_knowledge_base — search across all uploaded files and documents

Using with Claude Desktop

Add this server configuration to your Claude Desktop settings (Settings → Developer → MCP Servers), replacing the placeholder with your actual token:

{
  "mcpServers": {
    "linc-ai": {
      "url": "https://dashboard.withlinc.com/api/mcp",
      "headers": {
        "Authorization": "Bearer pmat_YOUR_TOKEN_HERE"
      }
    }
  }
}

Save and restart Claude Desktop — Linc AI should appear as an available MCP server.

Rate limits & security

• Rate limiting is applied per-token and per-IP.
• Audit logging records every MCP request — the tool called, outcome, and duration.
• Write operations require two-step confirmation — the client first requests the action, then confirms it.
• Token identifiers (prefix and last 4 characters) are shown in the Active Tokens list so you can track which token is in use.

Revoking tokens

Go to Settings → Developer → MCP Tokens, find the token in Active Tokens, and click Revoke. Revoked tokens are invalidated immediately. Rotate tokens periodically as a security best practice.