Tutorials

Build a Childcare Agent with CribScore

This guide shows the fastest path to an agent that answers childcare compliance questions without scraping public portals.

Recommended stack

Use remote MCP for the first pass. Move to the npm package when you want version pinning, local transport, or tighter environment control.

Questions it should answer

A useful childcare agent should be able to search by geography, compare providers, cite trust status, and quote evidence or events when they exist.

Remote MCP

{
  "cribscore": {
    "type": "streamable-http",
    "url": "https://api.cribscore.co/mcp",
    "headers": {
      "Authorization": "Bearer YOUR_API_KEY"
    }
  }
}

Decision Workflow

curl -X POST "https://api.cribscore.co/v1/decision/memo" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "geography": "90003",
    "child_age_months": 24,
    "budget_weekly": 350,
    "shortlist_limit": 3
  }'

Related guides

Remote MCP for Childcare Agents

Connect the hosted CribScore MCP endpoint to agent clients with one remote streamable HTTP config.

Local MCP Install with npm

Install the CribScore MCP package from npm and run it locally with `npx` or as a pinned dependency.

Replace State Portal Scraping

Use CribScore instead of building brittle childcare portal scrapers state by state.