Instrumentation Planning Skill

Instrumentation Planning

Strategic guidance for what to measure in web applications.

Core Question

For each user job, ask:

Did it complete? → completion rate
How long? → duration (p50, p95, p99)
What failed? → error type, context
Did they give up? → drop-off rate
How smooth? → Core Web Vitals, friction signals

Priority Tiers

| Tier | Focus | Priority | |------|-------|----------| | 1 | Errors + source maps | P0 - Day 1 | | 2 | Core Web Vitals | P0 - Day 1 | | 3 | User context + breadcrumbs | P0 - Week 1 | | 4 | Route transitions + API tracing | P1 - Week 2 | | 5 | Business metrics + synthetic tests | P2 - Month 1 |

Core Web Vitals Thresholds

| Metric | Good | Needs Improvement | Poor | |--------|------|-------------------|------| | LCP | ≤2.5s | ≤4.0s | >4.0s | | INP | ≤200ms | ≤500ms | >500ms | | CLS | ≤0.1 | ≤0.25 | >0.25 |

OTel-Compatible Naming

Use now for easier migration later:

http.request.method not method
http.request.duration not apiCallTime
url.path not page

Anti-Patterns

See references/anti-patterns.md for:

Measuring everything (noise, bundle bloat)
Skipping source maps
Blocking main thread
PII in breadcrumbs

Implementation Details

See references/instrumentation-patterns.md for:

Detailed 5-tier checklist
Span naming conventions
Sampling strategies

See references/jtbd.md for Jobs-to-be-Done framework.

Related Skills

skills/core-web-vitals - Tier 2 implementation
skills/error-tracking - Tier 1 implementation
skills/source-map-setup - Tier 1 implementation

Agent Skills: Instrumentation Planning

Install this agent skill to your local

Skill Files