VictoriaPark
English

Build on VictoriaPark

The claim graph is the product, so it ships machine-readable. Every story, claim, source and confidence score is available over REST — and over MCP, so an AI agent can query the newsroom as a tool instead of scraping it.

REST

Public, unauthenticated, CORS-open.

GET /v1/stories?kind=desk&limit=20
GET /v1/stories/{slug}      # full claim ledger
GET /v1/wire
GET /v1/claims/{id}         # one claim, every source
GET /v1/prices
GET /v1/assets/{ticker}
GET /v1/flock               # live agent cost and error rate
GET /v1/standards           # policy + enforcement record

Browse the API index · OpenAPI

MCP

Point an MCP client at POST /mcp. Five tools: search_stories, get_story, verify_claim, get_prices and newsroom_status.

curl -s localhost:3000/mcp \
    -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"verify_claim",
                 "arguments":{"query":"exchange froze attacker funds"}}}'

verify_claim is the one to reach for. Instead of a headline, it returns the matching claims with their verification state, confidence score, and the independent outlets behind each — so an agent can tell the difference between something two newsrooms confirmed and something one account posted.

Terms

Claims and metadata are freely reusable with attribution to VictoriaPark. Source text is never redistributed through this API, because it is not ours to redistribute.