Skip to content
Octarq

MCP server

octarq ships a built-in Model Context Protocol server, so an AI client — Claude Code, Claude Desktop, Cursor — can read your self-hosted company backend directly: “which links got the most clicks?”, “what mail landed today?”, “how many SaaS am I paying for?”

Running it

Section titled “Running it”

Run over stdio, the universal local MCP transport:

Terminal window
octarq mcp

For Claude Desktop, add to its MCP config:

{ "mcpServers": { "octarq": { "command": "/path/to/octarq", "args": ["mcp"] } } }

It reads the same .env / environment (and database) as the server.

Tools (all read-only)

Section titled “Tools (all read-only)”

list_links, list_mailboxes, list_emails, list_domains, export_data, and query_db_readonly — a general-purpose SQL tool so the AI can compute any metric without a bespoke endpoint.

Guardrails on query_db_readonly

Section titled “Guardrails on query_db_readonly”

This is the one place data reaches an LLM, so it is fenced:

  • Only a single SELECT / WITH runs, inside a read-only transaction — writes, PRAGMA, and ATTACH are rejected.
  • Results are row-capped.
  • Sensitive columns (password/token hashes, encrypted provider credentials, raw email bodies) are redacted.
  • Tools are scoped to one operator via OCTARQ_MCP_ORG_ID; for multi-tenant, run one octarq mcp process per tenant.