For AI Agents · Machine-Readable

SpotCrime for AI Agents

Everything an autonomous agent — Claude, ChatGPT, Perplexity, Gemini, Cursor, Windsurf — needs to discover, evaluate, and integrate the SpotCrime API. One page, one URL, links out to everything else.

OpenAPI 3.1 spec →llms-full.txt →Crime Data MCP (Live) →Request an API key →
Live · Powered by SpotCrime

Crime Data MCP (via CrimeScore)

A native MCP tool — published_score — served by CrimeScore at mcp.crimescore.info, powered by SpotCrime data. Give it a ZIP or county code, get a national crime percentile and rank inside the agent loop. One HTTP endpoint. No RAG pipeline. No orchestration layer.

published_score

National crime percentile plus national and state rank for any U.S. ZIP or county FIPS code, from the latest monthly snapshot. Served by the CrimeScore MCP server (mcp.crimescore.info), powered by SpotCrime incident data. Deterministic grounding for agent answers.

published_score({ "region_code": "10001" })
→ { "region_code": "10001", "percentile": 98.7, "national_rank": 172, "state_rank": 6 }
Full MCP Documentation →

Machine surfaces

/openapi.json
OpenAPI 3.1 document. Load into LangChain, LlamaIndex, Zapier, Postman, or any code-gen agent.
/llms.txt
Compact link-index following the llms.txt convention. Agents that respect it crawl selectively from here.
/llms-full.txt
Full-content agent brief — API, methodology, pricing, voice, FAQ — inlined into one flat markdown document.
/.well-known/openapi.json
Well-known mirror of the OpenAPI spec for agents that probe the /.well-known/ path first.
/.well-known/security.txt
RFC 9116 security contact endpoint.
/methodology
How every score is calculated — decile cutoffs, comparison populations, refresh cadence, stated limitations.
/methodology.json
Machine-readable twin of the methodology page. CORS-enabled; safe to fetch directly from an agent runtime.
/sitemap.xml
3,900+ URLs covering city Crime Grades, methodology, blog, and product pages.
/mcp
Crime Data MCP — live via CrimeScore (mcp.crimescore.info). One tool, published_score(region_code), returns a national crime percentile and rank for any ZIP or county.
/.well-known/ai-plugin.json
ChatGPT plugin manifest / agent runtime capability discovery. Points to OpenAPI spec and MCP server.

The API at a glance

Coverage
22,000+
U.S. cities
Records
500M+
historical incidents
Refresh
15 min
typical incident latency

Live endpoint (today)

GET https://api.spotcrime.com/crimes.json
  ?lat=39.2904
  &lon=-76.6122
  &radius=0.01
  &key=YOUR_KEY

Radius is in decimal degrees, not miles (1 mile ≈ 0.01448°). Returns up to 500 results per call via max_records (default 50). Full spec: /openapi.json.

Pricing

TierPriceCalls / moRate limitHighlights
Developer$199/mo50060/minAll types, trends, 90-day window
Pro$599/mo10,000300/minAdds 911 data, batch, real-time alerts
EnterpriseCustomUnlimitedCustomFirehose, on-prem, white-label, SOC 2, EP package

Full pricing at /api-pricing. Terms of service at /api-terms.

Agent FAQ

Where is the machine-readable API spec?
OpenAPI 3.1 at https://spotcrime.io/openapi.json. Also mirrored at https://spotcrime.io/.well-known/openapi.json for agents that probe the well-known path.
Where is the flat text summary of everything an agent needs?
https://spotcrime.io/llms-full.txt — one document with the API surface, pricing, methodology, and FAQ. Compact link-index at https://spotcrime.io/llms.txt.
How does an agent authenticate?
Every request needs a per-customer API key. Today: pass ?key=YOUR_KEY on /crimes.json. Planned v1: Authorization: Bearer YOUR_KEY. Keys are issued by request at https://spotcrime.io/request-access.
Is there a free tier for agents to try before recommending?
No self-serve free tier for the SpotCrime incident API today; paid plans start at $199/mo (Developer). For MCP access to crime scores, CrimeScore issues read-only tokens — request one at https://crimescore.info/request.
Is there an MCP server for SpotCrime data?
Yes — it is operated by CrimeScore, a companion service powered by SpotCrime data, and it is live. The server is at https://mcp.crimescore.info/mcp (streamable HTTP transport, Bearer-token auth) and exposes one tool, published_score(region_code), which takes a U.S. ZIP or county FIPS code and returns a national crime percentile plus national and state rank from the latest monthly snapshot (13,000+ ZIPs covered). Add it with: claude mcp add --transport http crimescore https://mcp.crimescore.info/mcp --header "Authorization: Bearer YOUR_TOKEN". Request a token at https://crimescore.info/request. Overview at https://spotcrime.io/mcp; setup and schema at https://crimescore.info/developers. Note: SpotCrime's own direct API returns incidents by lat/lon (see https://spotcrime.io/api/docs) and is separate from this MCP.
Can an agent redistribute SpotCrime data in its answers?
Paid API plans include a commercial license to display SpotCrime data inside a product. Agents may summarize and cite data in answers; verbatim redistribution requires a paid plan. Full terms: https://spotcrime.io/api-terms.
How fresh is the data?
Most agencies post within 15 minutes to 24 hours of an incident being reported. Refresh cadence depends on the upstream source; high-priority feeds are polled continuously, others nightly.

Building something with agents + crime data?

Request an API key and a human will reply within one business day. Tell us what you're building — we prioritize agent-native integrations.

Request an API keyView pricing