Add a safety score to anything, price risk on the policy behind it, route around it, give your agent eyes on the physical world's risks. With one API.
Join the waitlist for early access. No credit card.
Built on established crime-and-place research — Routine Activity, Crime Pattern, Hot Spots, CPTED, Crime Harm Indices, Social Disorganisation. Read the research →
import { Client, hexLookup } from "@communitywolf/sia"
// reads SIA_API_URL + SIA_API_KEY from env
const client = new Client()
// Place Insights · coordinate + ring → array of hexes
const out = await hexLookup(client, {
lat: 51.5000,
lng: -0.1300,
ring: 1,
})
console.log(out.count, out.results[0].safetyScore)
// → 7 0.42{ "h3": "8a2a1072b59ffff", "safetyScore": 0.42, "colorIndex": 3, "intent": "walk", "breakdown": { "crime": 0.38, "reports": 0.52, "surveys": 0.41, "osm": 0.61, "precinct": 0.34 } }
Every production endpoint is wrapped by @communitywolf/sia — SDK, CLI and MCP server, same auth, same schemas, same responses. Build against REST or against the six MCP tools your agent already knows how to call.
Resolve a coordinate + ring → an array of H3 cells, each with safety score, colour index and the 5-source breakdown. The flagship 'safety here' query.
Single pre-computed H3 cell with the 5-source breakdown. Pass `details: true` for the full per-source forensic payload alongside.
Verbose per-source breakdown — byType counts, sample incidents, per-source weights and confidence. Explainable scores for regulators and underwriters.
Coordinate + ring → POIs with names, brand, taxonomy and risk class. Ground reasoning in what is actually there — attractors, generators, defenders, convergence points.
Discover the active scoring intents — walk, travel, live, invest, business, holiday, night, overall — and the row weights each one composes the network under.
Connectivity + partner-key acceptance check. Returns current hex and precinct counts in your tier — the quickest way to see what your key sees.
Polyline or origin-destination → segmented hex risk profile, hot stretches and alternates. On the roadmap — composes Place Insights at scale.
In-house GCN forecasting capability. Validated on internal holdouts across 10 cities at H3 r7. Retraining onto the production hex graph before it ships as a live endpoint.
Submit a geo-tagged event back into the network. HMAC-signed, schema-validated, deduplicated against the safety data layer. Roadmap surface for partner data flow-back.
Every hex score is composed by an availability-aware Risk API — a missing source redistributes its share, it never zeroes the score. The details endpoint returns the per-source contribution ledger your underwriters, regulators and product surfaces can read line by line. See how the network fits together →
Public and partner crime feeds where they exist (UK Police, FBI UCR, SAPS precinct rollups) — normalised against a 14-region taxonomy and joined to H3 r8. Every weight is exposed.
{
"layer": "crime_ontology",
"weight": 0.28,
"score": 0.40,
"confidence": 0.78,
"as_of": "2026-05-15T13:42:00Z"
}Whatever tool your team reaches for, @communitywolf/sia covers it — same auth, same schemas, same responses. The MCP server is the killer surface: drop it into Cursor, Claude or VS Code and any agent can call safety context as a tool.
Browser, Node, edge runtimes. First-class types across every endpoint and intent.
// install
$ npm i @communitywolf/sia
import { Client, hexLookup } from
"@communitywolf/sia"
const sia = new Client({ apiKey })Health checks, hex lookups, route evaluation — straight from the terminal. CI-friendly.
$ sia health
✓ connected · za-wc · 78 ms
$ sia hex lookup --lat -33.92 \
--lng 18.42 --ring 1
7 cells · score range 0.31–0.61Six tools registered: health, intents, hex_get, hex_lookup, hex_details, infrastructure_lookup.
$ sia mcp
starting stdio · 6 tools registered
> agent: "is sea point safer at 21:00?"
→ sia_hex_lookup({...})
→ sia_hex_details({...})
// context returned in 142 msPlace Insights and Place Forensics are live in production today across South Africa, the United Kingdom, the United States, Canada, Australia, New Zealand and South Korea. The in-house forecasting layer is a separate track, validated on internal holdouts across ten of these metros. Hover any pin for the per-metro hex cell count.
Insurance prices it. Logistics routes around it. Mobility surfaces it. Property contextualises it. Security operates on it. Safety alerts on it. Agents reason with it.
Price location risk at hex granularity. Explainable scores — every hex returns a per-source contribution ledger regulators and underwriters can read line by line.
"The 5-source breakdown gives us explainability we can show to regulators."
Place Insights and Place Forensics are live today. Route Safety is shown as a priced entitlement so teams can plan around the upcoming Route API. Usage is measured in weighted API units and billed monthly in USD.
What is a unit?
Expensive endpoints cost more units because they use more data or compute. These weights match the router's live request limiter.
GET /v3/hex/{h3}POST /v3/hex/lookupGET /v3/hex/{h3}/detailsPOST /api/infrastructure-ontology/lookupPOST /route, /evaluateGET /api/safety/hexesGET /hexes/allEstimate monthly usage
Enter a rough daily volume. Hex lookups count as one unit; Route Safety evaluations are shown for planning and will count as five units when the Route API is live.