Discovery Glossary

Shadow Asset

An AI agent or resource found running in your infrastructure that is not instrumented with the Pavri SDK. Discovered via hosted-platform API polling.

Discovery Evidence

An immutable observation record from a discovery source. Contains: source identifier, three timestamps (observed_at, collected_at, processed_at), collection method, integrity hash, and the normalized asset data. Evidence records are append-only and never updated or deleted.

Correlation Candidate

A proposed link between a discovered asset and an SDK-registered agent. Carries a match type (exact, fuzzy, ambiguous), confidence score, and review state. Ambiguous candidates always require operator review.

Posture History

A timeline of changes to a discovered asset. Each entry records: what changed (state, tools, model, configuration), the previous and new states, and a link to the evidence that triggered the change.

Connector

A lightweight worker that polls a hosted AI platform API to discover deployed agents. Each connector implements: poll (fetch inventory), normalize (standardize records), and health check.

Evidence Integrity

The guarantee that all discovery evidence is immutable, carries provenance metadata, supports chain-of-custody linking via parent IDs, and enables deterministic replay of decisions.

Confidence Score

A numeric score (0.0–1.0) indicating how much trust to place in a piece of evidence, based on its source in the source-of-truth hierarchy.

Collection Method

How evidence was gathered: api_poll (scheduled platform query), webhook (push notification), scan (static analysis), or sdk_push (direct SDK telemetry).