Skip to main content
Conduit is designed to be called by LLM agents as much as by humans. Three things make it agent-native:
  1. /llms.txt — a compact, sectioned index of every endpoint (and /llms-full.txt for the full reference). Point an agent here first.
  2. OpenAPI 3.1 at /openapi.json — import into any codegen, or let an agent read it.
  3. An MCP server — gives your agent typed tools, with no glue code.

​
Discover the API as an agent

curl https://api.financialdatapi.com/llms.txt        # start here — sectioned endpoint index
curl https://api.financialdatapi.com/llms-full.txt   # full reference with parameters
curl https://api.financialdatapi.com/openapi.json    # OpenAPI 3.1

​
Add the MCP server

Install the Conduit MCP server and your agent gets tools like screen_macro, get_latest_observations, cross_country, rates_analytics, economic_calendar, and observation_provenance.
Claude Code
claude mcp add conduit \
  --env CONDUIT_API_KEY=your_key_here \
  -- node /absolute/path/to/conduit/mcp/dist/index.js
Once connected, an agent can answer macro questions by calling Conduit directly — for example, screening countries by inflation and unemployment, pulling a single indicator’s history, or tracing a value’s provenance — without you writing any integration code.

​
Conventions agents should know

  • Read-only: every endpoint is GET.
  • Envelope: { "data": ..., "meta": { "request_id", "api_version", "pagination" } }.
  • Pagination: cursor-based — follow meta.pagination.next_cursor while has_more is true.
  • Rights-aware: /v1/public/* returns only redistribution-safe data.
  • Stable error codes: bad_request, not_found, unauthorized, forbidden, rate_limited, internal_error.