DaedalMap
Docs / For Agents

The fastest way to understand
the agent API lane.

DaedalMap now exposes a live agent-facing discovery layer, a paid dataset-query lane, and an MCP endpoint. This page is the shortest path from “I found this randomly” to “I know how to make the first real call.”

Live now

Four flagship packs

The current published agent lane includes currency, earthquakes, volcanoes, and tsunamis.

Paid lane

HTTP 402 first

Paid data calls should challenge first. An unpaid request to the protected lane returns 402, then the client retries with payment.

MCP

Same execution path

The MCP server is a wrapper around the same underlying query and payment logic, not a separate data path with different rules.

What is live right now

The live public agent surface is built around free discovery plus paid execution. Discovery lives at /api/v1/guide, /api/v1/catalog, and /api/v1/packs/{pack_id}. Paid execution lives at /api/v1/query/dataset. MCP is available at /mcp.

What a first successful flow looks like

Start with discovery. Read the catalog, choose a pack, inspect that pack, and then make one narrow paid query. The first success pattern should be: free catalog, free pack detail, unpaid query returns 402, paid retry returns data.

Current flagship pack ids

The current published agent catalog is intentionally small and legible: currency, earthquakes, volcanoes, and tsunamis. Each pack maps to one or more executable source_id values underneath, but pack ids are the public discovery unit.

Paid lane mental model

The bought unit is a single normalized query against one underlying source. That keeps the first lane deterministic. You are not paying for a chat session or a hidden cross-pack planner. You are paying for one maintained queryable slice.

Best first call

If you just want to see the product shape, start with GET /api/v1/catalog, then GET /api/v1/packs/earthquakes, then a narrow paid query for one time window and a few regions.

What still matters before broader adoption

The runtime lane is live, but onboarding still matters. The next high-leverage improvements are worked examples, clearer public payment docs, MCP discovery metadata, and a stronger plain-language explanation of loc_id.