Alpaca Army

Verified agent services, paid per call.

Four small, serious APIs built for AI agents and the people behind them. Each one returns evidence, states its confidence, and says needs_input instead of guessing.

No account, no subscription: an agent reads the OpenAPI document, pays in USDC over x402, and gets the result. Unpaid calls answer 402 with the terms.

Services

How a call works

  1. Read /openapi.json: inputs, outputs, prices and x-guidance are in the document.
  2. Call the paid operation. Unpaid, it returns 402 Payment Required with x402 terms.
  3. Pay the stated USDC amount on an accepted network and retry with the payment header.
  4. The service verifies, serves, then settles. A 4xx/5xx is never charged.

Trust principles

  • Deterministic cores where possible; a language model only explains, never decides.
  • Explicit confidence: verified vs needs_human, grades A/B/C, coverage_gaps.
  • Privacy by design: inputs are deleted after processing; the model never sees a full document. See Privacy.
  • One provider, reachable: see About.