Connect your own AI agent (MCP)

Beyond the built-in assistant, Bilterm lets you connect your own AI agent — Claude, Grok, or any MCP-capable tool — directly to your portfolio data via the Model Context Protocol.

1. Go to Profile → AI Assistants & Copilots (“Connect authorized local or cloud-based assistants to securely manage properties under your account permissions”).
2. Click Create token. Name it after the integration (e.g. “Claude-desktop-read”), choose read-only or read-write scopes, and set an expiry. The token is shown once — copy it immediately.
3. Connect your agent:
   • Local: run the bilterm-mcp server on your machine with the token. The Profile page shows setup tabs for Claude Code, Grok, and Others.
   • Hosted: point a cloud agent at the hosted MCP endpoint with the token as the bearer credential.
4. Start with the whoami tool — it reveals exactly what the token can access.
5. Ask away: list properties, leases, tenants, rent payments, files, inbox conversations.

The 8 tools

whoami, list_properties, list_leases, get_lease, list_tenants, list_rent_payments, list_files, list_inbox_conversations.

Safety model

• The server holds zero permission logic — row-level security in the database decides every row. A restricted token returns empty results, not errors.
• Revoke a token on your Profile page and the agent’s access stops immediately.
• Every call is logged under Profile → Agent activity: which token, which tool, when.

Tips

• One token per agent. If something looks off, you revoke one integration, not everything.
• Start agents on read-only scopes. Grant write only when the agent needs to act, not just report.
• The tokens table shows each token’s prefix, expiry, last-used date, and scopes — review it the way you’d review who has keys to your buildings.