MCP serverMCP toolResearch Preview

Where to look first.

A probability heatmap of where a missing person is most likely to be found, built from their behavioural profile, the terrain, and current weather.

mcp.footstep.ai / predict
Why you need it

Why your agent needs this.

When the user asks the agent 'where might Mrs Patterson be?', the agent needs one tool that takes a behavioural profile, an area, and current conditions, and returns a usable probability surface. predict is that tool. Token-lean by default, render-ready by design. The agent picks it; you wrote none of the orchestration.

What it does for you

What your agent gets back.

  • Returns a render envelope with one H3 hex layer per probability band.
  • Probability scores conditioned on behavioural profile, terrain, and live weather.
  • Token-lean: full hex grid is opt-in via include_hexes.
  • Works in Claude Desktop, Cursor, Vercel AI SDK, OpenAI Agents, and any MCP-aware runtime.
  • Composes naturally with parse_address for radio reports and get_isochrone for reach analysis.
Try it

Plug it into your agent.

Add Footstep to your MCP-aware runtime. Your model picks the tool at runtime; you wrote none of the orchestration.

Prompt that triggers predict
"Generate a probability surface for a despondent missing person from the trailhead, 5km radius."
claude_desktop_config.json
{
  "mcpServers": {
    "footstep": {
      "url": "https://mcp.footstep.ai",
      "headers": {
        "x-api-key": "sk_live_your_key_here"
      }
    }
  }
}

Ready to call it?

£5 free credit on signup. One auth header, every major runtime.