agentleFS
Sn4kyGitSn4kyGit/qa-agent-blueprint/.claude/skills/perf-analysis/SKILL.md

perf-analysis

Skill2 starsChanged 7 days agoApache-2.0View on GitHub

Analyze session performance — duration, phase breakdown, token usage, bottlenecks, efficiency trends, and concrete improvement suggestions. Use for "/perf-analysis", "performance analysis", "Performance-Analyse", "warum ist das so langsam", "wie lange dauert ein Testlauf", "Token-Verbrauch analysieren", "Performance verbessern", "Kosten optimieren". Not for test quality metrics (/qa-report) or skill usage (/skill-review).

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the report. The summary table shows where you stand against targets.

    line 24

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/perf-analysis — how long does it take, what does it cost, and what should change

The blueprint writes timing and usage data for every run — and never reads it back. This turns that data into the numbers that answer "why is this slow?" and "what should we fix?" — with concrete, actionable suggestions.

Steps

  1. Run the analysis:
python3 .claude/skills/perf-analysis/perf-analysis.py --since 30 --suggest
# --project SHOP    one project only
# --ticket SHOP-123 one specific run
# --json            machine-readable output
  1. Read the report. The summary table shows where you stand against targets; the phase breakdown shows where time goes; the suggestions section names bottlenecks with specific fixes.
  2. Interpret against the project's gates (PROJECT.md → "Pipelines / quality gates") — if the project declares tighter targets, those win.
  3. Turn red/yellow items into actions. Each suggestion maps to a concrete skill or workflow change (see references/improvement-patterns.md for the full pattern library).
  4. Report the findings. For a team: use report-to-jira or the project's reporting format. For a quick look: just print the report.

What this skill never does

  • Never estimates — if a metric can't be computed from the data, it says "not computable" rather than guessing.
  • Never mutates — this is read-only. It doesn't change configs, tests, or project files. Improvement suggestions are proposals, not automatic fixes.
  • Never duplicates /qa-report numbers — pass rate, flake rate, and findings belong to qa-report. This skill focuses on time and cost.
  • Never reads transcripts — token usage and model affinity come from session-perf.log (Stop-hook-extracted metadata), never content.

Data sources

SourceWhat it provides
execution_summary.jsonverdict, duration_s, phases, session, test_cases
run.jsonduration_s, result counts
evidence/.audit/session-perf.logtoken usage per session + measured model affinity (model=, model_switches=)
evidence/.audit/compression.logmeasured compression bytes per seal (run, mode, original/compressed)
evals/results/history.logpass-rate trend

Fields not yet in older summaries are simply absent — the analysis reports what's available and notes what's missing.

Command surface

/perf-analysis [--since DAYS] [--project KEY] [--ticket KEY]

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.