Imported from gyrinx-app/gyrinx (
.agents/skills/trace-analysis/SKILL.md). Install upstream withnpx skills add gyrinx-app/gyrinx --skill trace-analysis. Copyright stays with the author.
Trace Performance Analysis
You are helping analyze OpenTelemetry/Google Cloud Trace files for performance optimization.
Playbook Context
@playbooks/trace/instructions.md
When to Use
This skill applies when:
- User wants to analyze a trace file for performance issues
- User mentions "trace playbook" or similar
- User has a
.jsontrace file from Google Cloud Trace - User wants to compare before/after traces
- User asks about N+1 queries, slow operations, or performance bottlenecks in a trace
How to Execute
-
Load the playbook instructions from
playbooks/trace/instructions.md -
Follow the stage-by-stage workflow:
- Stage 0: Setup (interactive) - validate trace, establish scope
- Stage 1: Per-Group Analysis (parallel) - analyze each operation group
- Stage 2: Aggregation - cluster patterns across groups
- Stage 3: Synthesis - impact analysis and prioritization
- Stage 4: Recommendations - generate actionable outputs
-
Use the provided templates in
playbooks/trace/templates/for consistent output -
Tag issues using
playbooks/trace/reference/issue-taxonomy.md -
Verify completion with
python playbooks/trace/scripts/verify-completion.py
Key Resources
| Resource | Path | Purpose |
|---|---|---|
| Master instructions | playbooks/trace/instructions.md |
Orchestration guide |
| Issue taxonomy | playbooks/trace/reference/issue-taxonomy.md |
Classification tags |
| Glossary | playbooks/trace/reference/glossary.md |
Term definitions |
| Templates | playbooks/trace/templates/*.md |
Output formats |
Example Interactions
User: "Analyze the trace file trace.json using the trace playbook" Action: Execute full playbook workflow, creating namespaced output directory
User: "Compare these two traces before.json and after.json" Action: Run playbook on both, then create comparison analysis
User: "What N+1 patterns are in this trace?" Action: Focus on Stage 1-2 to identify and aggregate N+1 patterns
Output Location
All analysis outputs go to playbooks/trace/output/{trace-id}/:
SUMMARY.md- Start here for executive summarygroups/- Per-operation analysisaggregation/- Pattern clustersrecommendations/- Action cards
Input traces are archived to playbooks/trace/input/{trace-id}/trace.json