I wrote the standard for making websites AI-operable. Learn More

Intelligence · API + MCP

Read the data programmatically.

JSON over HTTP for everything public on tylerewillis.com/intelligence. An MCP server for any client that speaks Model Context Protocol — Claude Desktop, Cursor, Zed, Continue, Cline, the rest. No key for public datasets. Paid key for leads + bulk + webhooks.

MCP 2024-11-05 OpenAPI 3.1 Free tier — no key

MCP Server Free preview

Streamable HTTP transport. One POST endpoint, JSON-RPC 2.0. Exposes the five intelligence datasets as resources plus two tools (search_intelligence, get_intelligence) for query-driven access. Read-only.

Endpointhttps://tylerewillis.com/intelligence/api/mcp
TransportHTTP POST, JSON-RPC 2.0
AuthNone (free preview)
Resources5 — one per dataset
Toolssearch_intelligence, get_intelligence

Claude Desktop / Cursor / Zed — config snippet

Drop into the MCP servers block of your client's config file. Most clients reload on save.

JSON
{
  "mcpServers": {
    "tylerewillis-intelligence": {
      "url": "https://tylerewillis.com/intelligence/api/mcp"
    }
  }
}

Verify by hand — JSON-RPC over curl

bash
curl -s https://tylerewillis.com/intelligence/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | jq

Call a tool

bash
curl -s https://tylerewillis.com/intelligence/api/mcp \
  -H 'Content-Type: application/json' \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": {
      "name": "search_intelligence",
      "arguments": { "query": "n8n", "dataset": "tools", "limit": 3 }
    }
  }'

Pre-baked prompts

Four prompts ship with real data pre-loaded into the conversation context. Surfaced in client UIs (Claude Desktop, Cursor, etc.) as one-click buttons. Almost no MCP server ships prompts yet — first-mover positioning.

PromptWhat it does
compare_toolsPulls full profiles for two tool slugs, frames a side-by-side comparison.
recommend_for_stackGiven a comma-separated tech stack, finds matching playbooks via the graph and frames an implementation pick.
weekly_digestInlines recent change_log events, asks the model to write a digest.
find_automation_forFree-text problem statement → semantic search across playbooks / case studies / tools → grounded recommendation.

REST Endpoints Free, no key

Plain HTTP GET, returns JSON. All public datasets. CORS open. Cache-Control sends short max-age so clients can revalidate every couple minutes.

MethodPathReturnsRows
GET /intelligence/api/tools.json Tools list (paginated) 116
GET /intelligence/api/tools/{slug}.json One tool, full payload
GET /intelligence/api/benchmarks.json Benchmarks list (paginated) 77
GET /intelligence/api/benchmarks/{slug}.json One benchmark, full payload
GET /intelligence/api/playbooks.json Playbooks list (paginated) 26
GET /intelligence/api/playbooks/{slug}.json One playbook, full payload
GET /intelligence/api/case-studies.json Case Studies list (paginated) 18
GET /intelligence/api/case-studies/{slug}.json One case studie, full payload
GET /intelligence/api/market.json Market Landscape list (paginated) 79
GET /intelligence/api/market/{slug}.json One market landscape, full payload
GET /intelligence/api/search?q=… Semantic search (vector similarity, all 5 datasets)
GET /intelligence/api/by-uuid/{uuid}.json Resolve any entity by its stable UUID (survives slug renames)
GET /intelligence/api/changes.json Recent changes — new entries + market status changes
GET /intelligence/api/openapi.json OpenAPI 3.1 specification
POST /intelligence/api/mcp MCP JSON-RPC endpoint

List with search + pagination

bash
curl -s 'https://tylerewillis.com/intelligence/api/tools.json?q=automation&limit=10&offset=0' | jq

Fetch one entity

bash
curl -s https://tylerewillis.com/intelligence/api/market/tools-n8n.json | jq

Semantic search

Vector similarity, not keyword. Describe the concept and get the closest entities by meaning. Each result includes a _score (cosine, 0..1).

bash

What changed recently

New entries across every dataset plus market status changes (acquisitions, shutdowns, pivots) from the last N days. Foundation for the "what changed in AI automation this week" feed.

bash
curl -s 'https://tylerewillis.com/intelligence/api/changes.json?since_days=7' | jq

List response shape

JSON
{
  "dataset":     "tools",
  "total":       116,
  "limit":       50,
  "offset":      0,
  "returned":    50,
  "next_offset": 50,
  "items": [ /* ... */ ],
  "docs":        "https://tylerewillis.com/intelligence/api"
}

Tiers

Public datasets are free. Leads, bulk export, and webhooks live behind a key. Issued by hand for now.

Free $0 / no key

The five public datasets. Soft cap, abuse-only enforcement.

  • All /intelligence/api/*.json endpoints
  • MCP server — full read access
  • ~60 req/min per IP (soft, just polite)
  • No SLA, but the data updates daily and the endpoints are CDN-cached

Pro Request access

Custom arrangements. Common asks below — tell me what you actually need.

  • Leads dataset (the subscription product) — API access
  • Webhooks on new entities
  • Bulk / mirror export
  • Higher rate limits + an SLA
  • Issued by hand — use the form below

Request a Pro key

Issued by hand right now. Tell me what you're building. I'll reply with the key and rate-limit terms within a day.