API & MCP

The procurement graph, for your agents.

Every screen in the product maps to an API call that returns the same data — clean JSON, documented in plain markdown, with curl examples. There's an llms.txt at the root and an MCP server so any AI agent can discover and use it without a human.

Get your API key Read the docs

request.sh
response.json
mcp.json
# Search opportunities matched to a business profile
curl https://api.geofabric.com/v1/opportunities/search \
  -H "Authorization: Bearer $GEOFABRIC_KEY" \
  -G \
  --data-urlencode "naics=541512" \
  --data-urlencode "set_aside=SDVOSB" \
  --data-urlencode "min_match=80"

Endpoints

One endpoint per question.

The same data behind the UI, exposed as a small, predictable REST surface. Cursor-paginated, rate-limited per plan, versioned under /v1.

GET/v1/opportunities/searchMatch opportunities to a NAICS, set-aside, or full business profile

GET/v1/entities/{uei}Full entity profile — contracts, officers, related companies, compliance

GET/v1/entities/{uei}/networkOfficer and related-entity graph with exclusion cross-checks

POST/v1/compliance/screenScreen a company or person against SAM + OIG LEIE exclusions

GET/v1/awards/searchHistorical & recent contract awards with recompete indicators

GET/v1/recompetesContracts approaching end of period of performance

POST/v1/askNatural-language query — returns a summary, structured data, and sources

Agent-ready by design

Built for humans and machines.

llms.txt at the root

A plain-text map of the API at geofabric.com/llms.txt, so any model can discover what’s available and how to call it — no scraping, no guesswork.

OpenAPI spec in YAML

A complete, versioned OpenAPI spec. Generate a typed client in your language of choice, or point your tooling straight at it.

Ambient-ready responses

Every response includes a plain-English summary alongside the structured data — so an answer makes sense read aloud, not just rendered in a table.

MCP server

Connect it to Claude, Cursor, or your own agent.

Point any MCP-compatible client at our server and your assistant can screen contractors, search opportunities, and pull entity profiles — with sources — inside the conversation. Same data, same freshness, no glue code.

Get your API key

mcp.json
# Add the GeoFabric MCP server to your client
# (claude desktop, cursor, or any MCP-compatible agent)

{
  "mcpServers": {
    "geofabric": {
      "url": "https://mcp.geofabric.com/sse",
      "headers": { "Authorization": "Bearer $GEOFABRIC_KEY" }
    }
  }
}

# Then just ask your agent:
#   "Is Meridian Atlas Holdings clean? Check the officers too."