Skip to main content

lading-optimize-hunt

Coordinates optimization attempts. Captures baselines, implements changes, invokes review, and records outcomes.

الانتقال إلى التثبيت

معلومات المصدر

المستودع
DataDog/lading
آخر نشاط في المصدر
٣ مارس ٢٠٢٦ في ١٧:٠١
لغة SKILL.md المكتشفة
الإنجليزية
النجوم
٩٨
التفرعات
١٥

خيارات التثبيت

يُحدَّد Prompt الذي يراجع المصدر أولًا بشكل افتراضي. يمكنك التبديل إلى أمر مباشر أو تنزيل نسخة محلية.

مراجعة ملفات المصدر

اقرأ SKILL.md وأي ملفات مرافقة يعرضها SkillsMP قبل أن تقرر التثبيت.

مستكشف الملفات
9 ملفات

عرض SKILL.md

SKILL.md
تعليمات المصدر · معاينة للقراءة فقط
name
lading-optimize-hunt
description
Coordinates optimization attempts. Captures baselines, implements changes, invokes review, and records outcomes.
allowed-tools
Bash(cat:*) Bash(cargo:*) Bash(ci/*:*) Bash(hyperfine:*) Bash(*/payloadtool:*) Bash(tee:*) Read Write Edit Glob Grep Skill
# Optimization Hunt Coordinates optimization attempts: captures baselines, implements changes, invokes review, and records all outcomes. ## Role: Coordinator and Recorder Hunt is the **coordinator and recorder** — it captures baselines, implements changes, hands off to review, and records all outcomes. Hunt does NOT: - Run post-change benchmarks (review does this) - Make pass/fail decisions on optimizations (review does this) Hunt DOES: - Record all verdicts and outcomes in `.claude/skills/lading-optimize-hunt/assets/db.yaml` after review returns --- ## Phase 0: Pre-flight Run `/lading-preflight`. --- ## Phase 1: Find Target Run `/lading-optimize-find-target`. It returns a YAML block with 6 fields: `pattern`, `technique`, `target`, `file`, `bench`, `fingerprint` - Print it out. --- ## Phase 2: Establish Baseline **CRITICAL: Capture baseline metrics BEFORE making any code changes.** ### Identify the Benchmark Target Use the `bench` and `fingerprint` fields from find-target's output — they are repo-relative paths ready to use: ```bash BENCH=<bench field without extension> # e.g. from "lading_payload/benches/syslog.rs" use "--bench syslog" PAYLOADTOOL_CONFIG=<fingerprint field> # e.g. "ci/fingerprints/syslog/lading.yaml" ``` ### Stage 1: Clear previous benchmarks Clear any previously captured baselines so stale data cannot contaminate this run. ```bash rm -f /tmp/criterion-baseline.log /tmp/baseline.json /tmp/baseline-mem.txt rm -rf target/criterion ``` ### Stage 2: Micro-benchmark Baseline Run **only** the benchmark for your target: ```bash cargo criterion --bench "$BENCH" 2>&1 | tee /tmp/criterion-baseline.log ``` ### Stage 3: Macro-benchmark Baseline Use the matching fingerprint config: ```bash cargo build --release --bin payloadtool hyperfine --warmup 3 --runs 30 --export-json /tmp/baseline.json \ "./target/release/payloadtool $PAYLOADTOOL_CONFIG" ./target/release/payloadtool "$PAYLOADTOOL_CONFIG" --memory-stats 2>&1 | tee /tmp/baseline-mem.txt ``` **Baseline captured.** These files will be consumed by review: - `/tmp/criterion-baseline.log` — micro-benchmark baseline - `/tmp/baseline.json` — macro-benchmark timing baseline - `/tmp/baseline-mem.txt` — macro-benchmark memory baseline **CRITICAL: All benchmarks must complete before continuing.** --- ## Phase 3: Implement Make ONE change. Keep it focused and minimal. Before proceeding, ALL changes must pass: ```bash ci/validate ``` **No exceptions. If ci/validate fails, fix the issue before continuing.** If `ci/validate` repeatedly fails on a **pre-existing bug** (not caused by your change), document it and stop. --- ## Phase 4: Hand Off to Review Run `/lading-optimize-review` with the target fields as positional arguments: ``` /lading-optimize-review <bench> <fingerprint> <file> <target> <technique> ``` Where: - `<bench>` — benchmark name from find-target's `bench` field, without path or extension (e.g. `trace_agent`) - `<fingerprint>` — repo-relative path from find-target's `fingerprint` field - `<file>` — repo-relative path from find-target's `file` field - `<target>` — function name from find-target's `target` field - `<technique>` — technique from find-target's `technique` field It returns a YAML report. Print it out. --- ## Phase 5: Recording After review returns its YAML report, record the result. Every outcome MUST be recorded. ### Step 1: Write the Report Write review's YAML report **verbatim** to `.claude/skills/lading-optimize-hunt/assets/db/<id>.yaml`. Do not modify, reformat, or add to the report content — it is the authoritative record from review. ### Step 2: Update the Index Add an entry to `.claude/skills/lading-optimize-hunt/assets/db.yaml` following the format in `.claude/skills/lading-optimize-hunt/assets/index.template.yaml`. ---
عرض على GitHub