The two embedding surfaces—API and widget—are designed to fit inside an existing earn or treasury product without forcing the partner to rebuild their UX.Documentation Index
Fetch the complete documentation index at: https://docs.pathprotocol.finance/llms.txt
Use this file to discover all available pages before exploring further.
API
Path’s API is a clean REST + GraphQL layer over the prediction stream, organised around four resource families:Pools
Pools
GET /v1/signals/pools returns the current prediction and
liquidity-impact context for every monitored pool, with
confidence-score filtering and per-protocol-family pagination.Predictions
Predictions
GET /v1/signals/predictions/calendar returns the time-series of
predictions for any single pool over a user-specified window.
Useful for back-testing a strategy against Path’s historical
output before going live.Briefing
Briefing
GET /v1/signals/predictions/briefing returns the daily summary—
top movers, regime flips, biggest accuracy wins of the last 24
hours. Designed to power a “Path daily” email or Slack post.Backtest
Backtest
GET /v1/signals/backtest returns the realised performance of a
hypothetical allocation strategy defined by the caller. Used by
partners to assess Path’s signal quality on their own historical
portfolio.Widget
The drop-in widget renders Path’s allocation intelligence inside an embedding surface. The widget reads the connected wallet (or an explicit balance from the partner), calls Path’s API, renders:- The current allocation breakdown
- The recommended re-balance with confidence + break-even
- The liquidity-impact context for any deposit size
- An expandable panel showing the signal source and accuracy band
Deployment
Hosted
Path operates the API. Partners hit
api.pathprotocol.finance/v1. Standard rate limits + SLA
documented per agreement.Self-host (Enterprise)
Partners can self-host the prediction stream behind their own
edge. The model artefacts ship as signed binaries; the partner
runs them inside their own infrastructure. Useful for partners
with strict data-residency requirements.