agentleFS
itsofficialadiitsofficialadi/boundiff/.agents/skills/performance-investigation/SKILL.md

performance-investigation

Skill0 starsChanged 25 days agoMITView on GitHub

Diagnose a measured latency, memory, throughput, or resource regression and propose the smallest supported optimization. Use for performance questions; measure before tuning.

9 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Files
SKILL.md

Performance investigation

A faster-looking implementation is not necessarily faster. Start with the metric and workload, not the optimization technique. Diagnosis is read-only by default.

  1. Define the symptom: p50/p95/p99 latency, throughput, CPU, memory, allocation, query count, or another measurable target. Identify representative inputs and hardware/environment when known.
  2. Establish a baseline using existing traces, profiles, metrics, or a reproducible benchmark. If you cannot measure, label the work as hypothesis generation.
  3. Narrow the hot path. Check algorithmic work, I/O, database access, serialization, memory pressure, caching, network, and concurrency only where evidence points.
  4. Compare candidate remedies by expected benefit, correctness risk, complexity, and blast radius. Avoid architecture changes justified by a microbenchmark alone.
  5. On explicit optimization requests, make one bounded change using $surgical-change; preserve correctness and measure the same workload again. Otherwise report the recommendation without edits.
  6. Record before/after conditions and noise. Do not extrapolate an isolated local benchmark to production SLOs.

Output

Metric and workload, baseline evidence, bottleneck, proposed or authorized change, post-change evidence if run, limits. Report measurements with units and sample conditions.

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.