XAIP Playground — Trust Before Agent Delegation
A read-only browser demo for seeing how historical trust evidence can change an agent's tool choice.
AI agents increasingly choose tools, services, and counterparties on your behalf. Today, many of those choices are blind: the agent sees a name and a description, but not much evidence about how that option has behaved before.
XAIP adds behavior-derived trust evidence before delegation. This playground shows the basic idea without posting receipts, creating keys, or changing any live score.
Blind vs XAIP Selection
This frontend selection logic is an educational playground simplification. It is not live /v1/select behavior and does not change routing, scoring, or production decisions.
Live Trust Evidence
The table below fetches the same public Trust API used by the main dashboard. The current public dataset is MCP-heavy because MCP was the first integration target.
| Server | Trust | Verdict | Receipts | Flags | Observed class | Observed verifiability | Observed settlement |
|---|---|---|---|---|---|---|---|
| fetching... | |||||||
Observed metadata is display-only and does not affect current scoring or /v1/select behavior.