Skip to main content

​
Overview

Use the Atlas API when source-backed civic data needs to move into another workflow: a newsroom research notebook, a nonprofit CRM, a local directory, a coverage dashboard, or an agent that needs structured context. Start with these guides when you need the model, workflow shape, and evidence patterns. Use the generated reference for exact parameters, request bodies, response schemas, and status codes.

REST API reference

Search endpoints, schemas, authentication requirements, and request examples.

​
The API in one picture

Atlas has four API surfaces that fit together: If you only need exact field names, open the generated reference. If you need to understand what the data means and how to use it without stripping evidence, start with the guides.

​
The core model

Atlas records are not just rows. A useful API response usually contains three ideas:
  • Actors — people, organizations, initiatives, campaigns, and events.
  • Context — places, issue areas, source types, source freshness, and related actors.
  • Evidence — source URLs, publications, dates, and extraction context that explain why Atlas knows something.
The evidence is not decoration. If you export a record, brief a team, feed an agent, or import a contact into another system, preserve the source trail that supports the claim you plan to use.

​
Choose the right starting point

Use place-first search when the question starts with a community:
Use entity search when the question starts with a name, type, keyword, or broader geography:
Use sources before reuse:
Use workspace APIs only when private team context, lists, briefs, watches, or review states are part of the workflow.

​
Authentication model

Many read-only directory endpoints are public. API keys and OAuth are for higher volume access, discovery runs, writes, workspace data, and MCP clients. For most scripts and server-side integrations, use an API key. For compliant MCP clients and assistant workflows, use the OAuth flow the client starts from the MCP server URL. Read authentication before building anything that writes data, starts discovery, or accesses a workspace.

​
What not to copy into your integration

  • Do not treat a matching record as a verified recommendation without reading sources.
  • Do not drop source URLs or extraction context when exporting records.
  • Do not publish private workspace notes as public evidence.
  • Do not assume a discovery run result is ready for outreach just because the run completed.
  • Do not duplicate the generated reference in your own docs; link to it for exact shapes.

​
Next steps

Tutorial: make your first call

Follow a short place-first workflow from search to source review.

Find source-backed actors

Search entities and preserve evidence in downstream tools.

Understand place data

Use profiles, coverage, signals, and source lists together.

Observe civic signals

Query snapshots, open streams, and keep protocol contracts in sync.

Generated reference

Check exact endpoint parameters and response schemas.