メニュー
Run lightweight performance/load tests (QPS/latency) against project HTTP endpoints using hey, track regressions, and suggest optimizations. Use when the user asks to benchmark, load test, or measure performance after changes.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Generate and complete reusable security test documents under docs/security/ based on the current project implementation and/or a confirmed plan. Use when a developer asks to complete security test docs, add missing security scenarios, or establish a security test baseline for the project.
Generate and maintain reusable UI/UX test documents under docs/uiux/ based on the current project UI implementation and the design-system constraints. Use when a developer asks to complete UI/UX test docs, add missing UI/UX scenarios, align UI constraints, or wants UI consistency/a11y/responsive regression checks during development.
Guide for working with the Agent Orchestrator — a CLI tool for AI-native SDLC automation. Use when writing or editing YAML manifests (Workspace, Agent, Workflow, StepTemplate, ExecutionProfile, SecretStore, EnvStore, Trigger, CRD), running orchestrator CLI commands, designing workflow step pipelines, writing CEL prehook/finalize expressions, configuring triggers (cron/event-driven task creation), or configuring self-bootstrap workflows. Triggers: any mention of orchestrator config, YAML manifests with "orchestrator.dev/v2", workflow steps, prehooks, finalize rules, task create/start/pause, orchestrator run, step filtering, direct assembly, agent capabilities, StepTemplate prompts, execution profiles, sandbox configuration, secret management, or trigger/cron scheduling.
Govern feature request (FR) documents through their full lifecycle — from planning to implementation to closure. Use when the user asks to govern/治理 a feature request, close an FR, check FR status, or says "治理FR", "/fr-governance". Scans docs/feature_request/ for open FR docs, plans implementation, executes governance, and self-checks closure.
Generate or update QA/security/UIUX test documentation after confirmed feature implementation plans or completed refactors. Use this skill AFTER plan approval or code completion to: (1) add new QA test docs for new behavior, (2) generate design docs, and (3) run repo-wide cross-doc impact analysis (docs/qa/, docs/security/, docs/uiux/, docs/guide/ and its translations, docs/design_doc/, docs/showcases/, root Markdown including CHANGELOG.md, and .claude/skills/) to update stale steps, expectations, and assertions. Triggers when users ask to create QA docs, update test docs after implementation, or sync QA/security/UIUX docs after behavior changes.
Govern and periodically remediate QA documentation quality across docs/qa, docs/security, and docs/uiux. Use when users ask to audit QA docs, run recurring documentation checks, fix doc drift after refactors, enforce scenario limits/checklists/UI entry visibility, or synchronize README/manifest/changelog consistency.
| name | performance-testing |
| description | Run lightweight performance/load tests (QPS/latency) against project HTTP endpoints using hey, track regressions, and suggest optimizations. Use when the user asks to benchmark, load test, or measure performance after changes. |
Use hey to measure throughput and latency for key endpoints. This skill is project-agnostic and works best when the project exposes health/ready endpoints and one or more critical API endpoints.
hey installed (or installable)which hey || brew install hey
Start the environment (project-bootstrap default):
docker compose -f docker/docker-compose.yml up -d --build
If the backend is Rust and you are running it outside Docker, run in release mode:
cd core
cargo run --release
If the project has a scripts/benchmark.sh, use it:
# Quick test (lower concurrency range)
./scripts/benchmark.sh
# Full test (higher concurrency range)
./scripts/benchmark.sh full
# Custom target URL
BASE_URL=http://your-server:8080 ./scripts/benchmark.sh
Otherwise, run hey directly. Set the target base URL:
BASE_URL=http://localhost:8080
Smoke benchmark on a cheap endpoint (prefer /health if present):
hey -n 2000 -c 50 "$BASE_URL/health"
Benchmark a representative API endpoint (replace path and method/body):
hey -n 2000 -c 50 -m GET "$BASE_URL/api/v1/items"
POST example:
hey -n 1000 -c 25 -m POST -H "Content-Type: application/json" \
-d '{"name":"example"}' \
"$BASE_URL/api/v1/items"
| Endpoint | Purpose | Measures |
|---|---|---|
/health | Pure compute | Framework baseline QPS |
/ready | DB + Cache | I/O bound performance |
/api/v1/{resource} | Business logic | Real API performance |
| Rating | /health QPS | Notes |
|---|---|---|
| Excellent | > 30,000 | Optimal performance |
| Good | > 10,000 | Production ready |
| Fair | > 5,000 | Check bottlenecks |
| Low | < 5,000 | Likely debug mode |
Track:
If numbers regress after a change:
./scripts/reset-docker.sh if stateful)| Issue | Solution |
|---|---|
| Low QPS | Ensure --release flag (Rust), check resource usage |
| Service not running | Start via Docker Compose or cargo run --release |
| DB/Cache issues | docker compose ps, restart services |