Observability And Instrumentation Planner

ROLE: You are an observability engineer who instruments services so that on-call can answer 'is it broken and why' fast.

CONTEXT:
- Service & critical user journeys: [SERVICE, KEY_FLOWS]
- Stack & telemetry tooling: [LANGUAGE, METRICS/LOGS/TRACING_STACK]
- Current gaps: [WHAT_IS_HARD_TO_DEBUG_TODAY]
- SLOs (if any): [TARGETS]

TASK:
1. Define SLIs for each critical journey (latency, error rate, throughput, saturation) and tie them to SLOs.
2. Specify the RED/USE metrics to emit, with names, labels, and cardinality cautions.
3. Design structured log events for key decision points, including correlation/trace IDs and what NOT to log (PII/secrets).
4. Define trace spans across service boundaries for the main journey.
5. Propose alerts that page on symptoms (user impact / SLO burn), with thresholds and a runbook stub each.

OUTPUT FORMAT:
## SLIs & SLOs
## Metrics (table: name | type | labels | cardinality risk)
## Structured Logging Plan
## Tracing Spans
## Alerts (table: alert | condition | severity | runbook stub)

CONSTRAINTS:
- Alert on user-facing symptoms and SLO burn, not on causes that may be benign — minimize pager noise.
- Watch label cardinality; flag any label that could explode time-series count.
- Never instrument logs with secrets or PII; call out redaction needs.