observability-bootstrap (v2.0)

Bootstrap practical, low-noise observability so teams can detect, diagnose, and resolve incidents quickly.

Trigger Signals

Use when user asks to:

• add observability to a service/system
• improve telemetry quality (logs/metrics/traces)
• define baseline alerting and SLI/SLOs
• remove production blind spots

Modes

• audit-only (default): assess current telemetry and gaps
• bootstrap-plan: implementation plan + starter artifacts
• patch-safe: apply low-risk instrumentation/config updates (with confirmation)

Inputs

• service type (api, worker, webhook, frontend, mixed)
• environment tiers (dev/stage/prod)
• incident priorities / reliability targets (if available)

Coverage Model

1. Logs: structured, correlated, redacted
2. Metrics: latency, throughput, errors, saturation
3. Traces: end-to-end request/dependency visibility
4. Alerts: actionable and owner-mapped
5. SLI/SLO: measurable reliability objectives + error budget

Workflow

1. Identify top critical user/API/background flows.
2. Audit current telemetry pipeline and instrumentation coverage.
3. Score maturity and map high-impact gaps.
4. Define minimum viable observability baseline per flow.
5. Draft SLI/SLO and alert policy with ownership/runbook mapping.
6. Produce rollout plan (Phase 1/2/3) + verification checklist.

Minimum Baseline Checklist

• [ ] Request/trace IDs propagated end-to-end
• [ ] Structured logs include service/env/severity/correlation fields
• [ ] PII/secrets redaction policy enforced
• [ ] Golden signals captured for critical services
• [ ] At least one SLI/SLO per critical flow
• [ ] Alerts contain owner + runbook + clear trigger/resolution hints
• [ ] Dashboard exists for top 3 critical flows

Output Contract

Return exactly:

1. Observability maturity summary
2. Gap table (priority | area | current | missing | impact | fix)
3. Phase rollout plan (1/2/3)
4. SLI/SLO + alert starter recommendations
5. Verification checklist + next actions

Fail Conditions

Mark as not-operationally-ready if:

• critical flow has no alertable error signal
• no correlation ID/trace propagation on key paths
• no ownership/runbook mapping for high-severity alerts

Templates

Use:

• templates/sli-slo-template.md
• templates/alert-policy-template.md

Guardrails

• Prefer vendor-neutral standards unless user requests specific stack
• Balance visibility with telemetry cost (sampling/retention guidance)
• Never log secrets or sensitive personal data
• Optimize for actionable alerts, not alert volume