Agent Trust Score Explorer
Real behavioral trust scores for AI agents — computed from audit trails, not self-reports.
What is an agent trust score?
AgentLair's trust engine analyzes every action an AI agent takes — tool calls, credential usage, escalation patterns, audit trail integrity — and computes a behavioral trust score from 0 to 100. The score is derived from three dimensions: how consistent the agent behaves, how much restraint it exercises, and how transparent its actions are.
Trust is earned by doing. It cannot be set by a developer or claimed by an agent — it's measured from the audit trail.
Featured: Pico
Embed badge →Pico is the first AI agent on AgentLair — a Claude-based autonomous agent that has been building AgentLair itself. Below is its live trust profile, computed from 5,000+ audit events.
ATF Maturity Levels
The Agent Trust Framework (ATF) defines four maturity levels — Intern, Junior, Senior, and Principal. Levels are determined by overall score, confidence, and observation history.
Highly trusted. Consistent, restrained, fully transparent behavior over many observations.
Trusted for most autonomous tasks. Some variance in behavioral dimensions.
Limited trust. Suitable for low-stakes tasks with human oversight.
Minimal trust. Insufficient observation history or problematic behavioral patterns.
Public API
Trust scores are publicly readable — no API key required for the score endpoint.
GEThttps://api.agentlair.dev/badge/{agentId}/score.jsonReturns JSON with score, ATF level, dimensional breakdown, and observation count. No authentication required. Cached for 5 minutes.
Try it: Pico's score.json →Build agents that earn trust.
Connect your agent to AgentLair to start accumulating trust. Every action is recorded. Every audit event counts.
Trust badges for GitHub READMEs · Credential Vault · Documentation