Observability

0x0 production services emit structured operational evidence. For Live

applications, observability is part of the framework surface, not an optional

debug convention.

Live Failure Correlation

Every failing Live session report must carry the same identifiers through logs,

metrics, traces, and debug evidence:

• session-id;
• request-id;
• event-id;
• trace-id;
• route.

The source-owned policy lives at frameworks/live/observability-policy.tsv.

The bounded gate is:

make live-observability-check

The gate proves a failing session includes route, event, diff, and transport

trace spans; patch size, queue depth, and reconnect count metrics; a stable

error class; template and patch debug annotations; and production redaction.

Development And Production Modes

Development mode may include template and patch source annotations. Production

mode must keep structured output redacted and correlation-safe. The

observability gate rejects any policy row that permits secret leakage.

Diagnostics

Stable diagnostics are documented in docs/diagnostics.html under Live

Observability Diagnostics. Missing policy rows, missing correlation, incomplete

metrics, incomplete traces, incomplete debug annotations, incomplete modes, and

unsafe secret policy all fail closed.

Release Evidence

The Live framework ledger records observability evidence in

release/live-framework-surface.tsv. Final release discipline includes

make live-observability-check.