A contract and SDK for privacy‑conscious AI agent governance telemetry
AgentTrust Telemetry defines a backend‑neutral contract for capturing policy decisions, approvals, usage, data flows, and evidence lifecycle events in AI agent runtimes. It ships Python and TypeScript SDKs that integrate with OpenTelemetry, allowing developers to emit structured, privacy‑preserving telemetry without installing a full tracing backend. The library is useful for teams building autonomous agents who need auditable governance and cost tracking. Compared to ad‑hoc logging, it offers a formal schema, conformance tests, and a reusable contract that can be shared across projects.
View on GitHub →agentrust-io/agentrust-telemetry