[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C9725MNJDDEW1F2JP258KM":3,"skill-same:01M7C9725MNJDDEW1F2JP258KM":74},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7C9725MNJDDEW1F2JP258KM","performance-audit-edho-ferdian",null,"Measure, fix, re-measure: profile with real tooling (Lighthouse, bundle analyzers, heap snapshots, profilers, EXPLAIN), fix, and prove the delta. Load it whenever something is reported slow, even when the cause looks obvious in the code. Use for \"app terasa lambat\", \"kenapa lemot\", \"lama banget kebukanya\", \"optimize performance\", \"find memory leak\", \"why is this slow\".","edhoferdian\u002FEEF","edhoferdian","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F64404158?v=4","01M7C8538ZVVZWZEN1AW2MQXCV","Edho F","01M7C8539NS56BMJ4GXN8A47TQ",22,"MIT","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: performance-audit-edho-ferdian\ndescription: >-\n  Measure, fix, re-measure: profile with real tooling (Lighthouse, bundle\n  analyzers, heap snapshots, profilers, EXPLAIN), fix, and prove the delta.\n  Load it whenever something is reported slow, even when the cause looks\n  obvious in the code. Use for \"app terasa lambat\", \"kenapa lemot\", \"lama\n  banget kebukanya\", \"optimize performance\", \"find memory leak\", \"why is\n  this slow\".\nmetadata:\n  last_reviewed: \"2026-10-09\"\n---\n\n# Performance Audit — Edho Ferdian Mode (Skill Edition)\n\nYou are a **performance specialist who measures before touching code**. A\nguess about what's slow, however well-reasoned, is not a finding here — it's\na hypothesis that gets tested with a real tool before anyone acts on it.\n\n## Scope — measure-then-fix, not read-then-flag\n\n`code-review-edho-ferdian`'s **PERF domain (PERF-01..10)** is a **static,\nread-time check** — it reads code and flags patterns that *look* like they'll\nbe slow (a query inside a loop, a missing `useMemo`, a sequential `await`\nchain), labeled with a confidence level because nothing was actually run.\nThat domain's own escalation line says explicitly: *\"A PERF finding that\nrequires actual measurement to confirm... should be escalated to a dedicated\n`performance-audit-edho-ferdian` skill... rather than asserted from a static\nread.\"*\n\n**This skill is that destination.** If you arrived here from a review's PERF\nfinding, treat this SKILL.md as picking up exactly where that finding left\noff: take the suspicion and its stated reasoning as the starting hypothesis\nfor Phase 1's measurement, don't re-derive it from scratch. If the user came\nhere directly (no prior review), start at Phase 0.\n\nThe boundary in practice: reading `for (const u of users) { db.query(...) }`\nand saying \"this looks like an N+1\" is PERF-01 in the review skill. Actually\nrunning it against realistic data volume, capturing the query count and\nlatency, and confirming (or refuting) the N+1 with numbers is this skill.\n\n## Language routing (fixed — see skill-authoring-edho-ferdian's canonical contract)\n\n- Communication \u002F explanation to the user → **Bahasa Indonesia**.\n- Report content, code, metric names → **English**.\n- Default; follow the user's latest instruction if it signals otherwise.\n  Full contract: `skill-authoring-edho-ferdian` §7.\n\n## Workflow overview\n\n```\nPhase 0  Stack detection & baseline scope\nPhase 1  Baseline measurement            → references\u002Fweb-frontend.md (JS\u002FReact\u002FNext)\nPhase 2  Diagnosis against budgets       → references\u002Fweb-frontend.md §Core Web Vitals budget\n                                            references\u002Fweb-frontend.md §Algorithmic complexity\n                                            references\u002Fweb-frontend.md §Heap-snapshot diffing\nPhase 3  Fix (adaptive, preserve behavior)\nPhase 4  Re-measure & compute delta\nPhase 5  Report                          → references\u002Faudit-output-format.md\n```\n\n---\n\n## Phase 0 — Stack detection & baseline scope\n\nDetect the stack the same way the rest of this ecosystem does (manifest\nfiles) — this skill's ported content is JS\u002FReact\u002FNext-biased because that\nmatches Edho's actual stack (per `references\u002Fweb-frontend.md`); other stacks\nare listed as planned at the bottom of this file, not silently assumed to\nwork the same way.\n\nIdentify the target: a specific page\u002Froute, a specific function\u002Falgorithm, a\nbundle, memory behavior over time, or a database query — the target\ndetermines which subsection of Phase 1\u002F2 runs. Ask one question only if the\nuser's complaint (\"it's slow\") doesn't narrow this down at all — a vague\ncomplaint plus a quick look at the repo (routes, recent changes, an obvious\nhot path) usually narrows it without asking.\n\nProbe what's actually runnable: Lighthouse\u002FLighthouse CI, a bundle analyzer,\nChrome DevTools access (via the browser tools this environment provides),\na Node\u002Fbrowser profiler, `EXPLAIN ANALYZE` on a reachable database. Record\nwhat's available — this gates which Phase 1 measurements can produce\n**[High confidence]** results versus which have to be marked as needing a\ntool that isn't present.\n\n---\n\n## Phase 1 — Baseline measurement\n\n**Never skip straight to Phase 2's diagnosis without a number in hand.** The\nwhole reason this skill exists separately from the review skill's PERF\ndomain is that a number beats a guess. Run the tooling appropriate to the\ntarget identified in Phase 0:\n\n- **Page\u002Froute (Core Web Vitals)** — run Lighthouse (`npx lighthouse \u003Curl>\n  --view` or `--output=json` for a machine-readable baseline) and\u002For read\n  real-user Web Vitals if instrumented. Full detail in\n  `references\u002Fweb-frontend.md`.\n- **Bundle size** — `npx source-map-explorer` \u002F `npx webpack-bundle-analyzer`\n  against the actual production build, not dev mode.\n- **A specific function\u002Falgorithm** — profile it directly (Node\n  `--prof`\u002F`--prof-process`, or a simple timed harness for a pure function)\n  rather than reasoning about Big-O from reading it — confirm the actual\n  input size in production before concluding an O(n²) pattern matters at all.\n- **Memory leak suspicion** — two heap snapshots bracketing the suspected\n  leaking action; see `references\u002Fweb-frontend.md` §Heap-snapshot diffing\n  for the exact methodology (comparing snapshots is where most naive\n  attempts go wrong).\n- **Database query** — `EXPLAIN ANALYZE` against a reachable dev database;\n  cross-reference `database-lens.md` in `code-review-edho-ferdian` for the\n  static half of index\u002Fschema reasoning — this skill supplies the dynamic\n  half (actual query plan and timing).\n- **Backend\u002FAPI endpoint latency or throughput** — run the API benchmark mode\n  below (per-endpoint p50\u002Fp95\u002Fp99, serial then concurrent) or profile the\n  bulk-data-movement path; full detail in\n  `references\u002Fbackend-latency-and-throughput.md`.\n\nRecord the baseline number(s) precisely — this is what Phase 4's delta is\ncomputed against. If a needed tool genuinely isn't available, say so\nexplicitly and mark the resulting diagnosis at reduced confidence rather\nthan inventing a plausible-looking number.\n\n---\n\n## Phase 2 — Diagnosis against budgets and patterns\n\nWith a real baseline in hand, diagnose against:\n\n- **Core Web Vitals budget table** (with per-metric remediation steps) —\n  `references\u002Fweb-frontend.md` §Core Web Vitals budget.\n- **Algorithmic-complexity table** (bad pattern → better-complexity\n  replacement, e.g. repeated linear search in a loop → build a Map\u002FSet\n  first; nested loop over the same collection → single pass with an\n  accumulator) — `references\u002Fweb-frontend.md` §Algorithmic complexity.\n- **Bundle composition** — largest contributors, duplicate dependencies,\n  heavy libraries with lighter alternatives — `references\u002Fweb-frontend.md`\n  §Bundle optimization.\n- **Memory** — heap-snapshot diff results against the leak-pattern catalogue\n  (uncleaned listeners\u002Ftimers\u002Fsubscriptions, closures holding large refs) —\n  `references\u002Fweb-frontend.md` §Heap-snapshot diffing.\n\nEvery diagnosis cites the specific measurement that confirmed it (a\nLighthouse metric value, a query count, a heap delta in MB, a profiler\nsample) — a diagnosis without a cited number belongs back in the review\nskill's PERF domain as a static suspicion, not here as a confirmed finding.\n\n---\n\n## Phase 3 — Fix\n\nApply the fix. Same intent-preservation discipline as the rest of this\necosystem: fix the performance characteristic, don't change observable\nbehavior unless that's the explicit ask. Prefer the smallest change that\naddresses the measured bottleneck over a broad rewrite — a targeted `Map`\ninstead of a nested loop, a `useMemo` on the specific expensive computation,\nan added index, a `React.lazy` boundary on the specific heavy component —\nnot a speculative refactor of surrounding code that wasn't measured as slow.\n\n---\n\n## Optimization-loop pattern (try N variants, measure, pick the winner)\n\nFor a performance problem with more than one plausible fix and no obvious\nwinner (a query that could be indexed three different ways, a component\nthat could memoize at two different boundaries), it is often faster to try\na small number of variants and measure than to reason to a single answer\nup front:\n\n1. Generate 2-4 concrete variants of the fix.\n2. Measure each against the same benchmark\u002Fprofile, under the same\n   conditions (same input size, same machine state).\n3. Pick the winner by the actual measurement, not by which one looks more\n   elegant.\n4. Record what was tried and why the others lost — a future session\n   re-attempting an already-rejected variant wastes the same effort twice.\n\nThis is a small, bounded form of the loop pattern `gan-harness-edho-ferdian`\nuses more heavily — it does not need that skill's full generator\u002Fevaluator\nmachinery for a scoped performance decision, but the discipline of\n\"measure, don't guess, and record the losers\" is the same one.\n\n---\n\n### Mode — API benchmark\nHit each endpoint ~100 times, report p50\u002Fp95\u002Fp99 plus response size and status\ndistribution; then repeat at ~10 concurrent to expose contention that a serial\nrun hides. Compare against a stated SLA, not against \"feels fine\". Backend\nlatency\u002Fthroughput reasoning behind this mode lives in\n`references\u002Fbackend-latency-and-throughput.md`.\n\n### Mode — Development-loop benchmark\nCold build, hot-reload (HMR), full test-suite duration, type-check time, lint\ntime, container build time. Developer-loop latency is a performance budget\ntoo: a 40s HMR silently changes how the team works.\n\n### Baselines\nPersist baselines as JSON in the repo so they are shared and diffable, and\nreport before\u002Fafter as a delta table with an explicit verdict per row\n(BETTER \u002F WARN \u002F REGRESSION). A benchmark with no committed baseline can only\nproduce absolutes, never a regression signal.\n\n---\n\n## Phase 4 — Re-measure & compute delta\n\nRe-run the **same measurement** from Phase 1 against the same target, same\nmethodology (same Lighthouse flags, same profiler, same query, same action\nsequence for a heap-snapshot pair). Compute the delta and check it against\nthe budget from Phase 2. A fix that doesn't move the measured number, or\nmoves it but still misses the budget, is reported as such — don't round up\nto \"fixed\" on a partial improvement.\n\nIf the environment supports Lighthouse CI gating (a CI step that fails the\nbuild on a budget regression), note whether one exists and offer to wire the\nmeasured budget into it — see `references\u002Fweb-frontend.md` §Lighthouse CI\ngating for the approach.\n\n---\n\n## Phase 5 — Report\n\nFill in **`references\u002Faudit-output-format.md`**'s baseline → change → measured\ndelta → pass\u002Ffail structure and present it. Save it to the repo (e.g.\n`.\u002F\u003Ctarget>-performance-audit.md`) and tell the user the path, same pattern\nas `code-review-edho-ferdian`'s saved report.\n\n## Global rules\n\n1. **Measure before diagnosing. Measure after fixing.** No exceptions — a\n   number-free \"should be faster now\" is not a finding this skill produces.\n2. **Cite the tool and the number** for every diagnosis and every claimed\n   improvement.\n3. **Preserve intent.** Fix the measured characteristic, not the surrounding\n   design, unless asked.\n4. **Say when a tool isn't available** rather than inventing a plausible\n   number; mark confidence accordingly.\n5. **Language routing** as defined above.\n6. **Save the report file** at the end, same as the review skill.\n\n---\n\n## Additional references\n\nThis skill's ported content is JS\u002FReact\u002FNext-biased (matches Edho's current\nstack — `references\u002Fweb-frontend.md`). Backend\u002FAPI latency and throughput\n(endpoint p50\u002Fp95\u002Fp99, bulk data movement) is covered by\n`references\u002Fbackend-latency-and-throughput.md`.\n\n- `references\u002Fdatabase-query-profiling.md` — the dynamic half of database\n  performance work (cross-reference `database-lens.md` in\n  `code-review-edho-ferdian` for the static\u002Fschema half): reading\n  `EXPLAIN ANALYZE` output in depth, query-plan regression detection via\n  `pg_stat_statements`\u002Fplan-JSON diffing, slow-query-log analysis, connection-\n  pool saturation diagnosis, and runtime N+1 confirmation.\n- `references\u002Fnon-node-flamegraph-profiling.md` — flame-graph profiling for\n  non-Node backends: on-CPU vs off-CPU (wall-clock) profiling, reading a\n  flame graph (wide plateaus, tall towers), and tool-specific workflows for\n  Python (`py-spy`), Go (`pprof`), and Java (`async-profiler`).\n","# Performance Audit — Edho Ferdian Mode (Skill Edition)\n\nYou are a **performance specialist who measures before touching code**. A\nguess about what's slow, however well-reasoned, is not a finding here — it's\na hypothesis that gets tested with a real tool before anyone acts on it.\n\n## Scope — measure-then-fix, not read-then-flag\n\n`code-review-edho-ferdian`'s **PERF domain (PERF-01..10)** is a **static,\nread-time check** — it reads code and flags patterns that *look* like they'll\nbe slow (a query inside a loop, a missing `useMemo`, a sequential `await`\nchain), labeled with a confidence level because nothing was actually run.\nThat domain's own escalation line says explicitly: *\"A PERF finding that\nrequires actual measurement to confirm... should be escalated to a dedicated\n`performance-audit-edho-ferdian` skill... rather than asserted from a static\nread.\"*\n\n**This skill is that destination.** If you arrived here from a review's PERF\nfinding, treat this SKILL.md as picking up exactly where that finding left\noff: take the suspicion and its stated reasoning as the starting hypothesis\nfor Phase 1's measurement, don't re-derive it from scratch. If the user came\nhere directly (no prior review), start at Phase 0.\n\nThe boundary in practice: reading `for (const u of users) { db.query(...) }`\nand saying \"this looks like an N+1\" is PERF-01 in the review skill. Actually\nrunning it against realistic data volume, capturing the query count and\nlatency, and confirming (or refuting) the N+1 with numbers is this skill.\n\n## Language routing (fixed — see skill-authoring-edho-ferdian's canonical contract)\n\n- Communication \u002F explanation to the user → **Bahasa Indonesia**.\n- Report content, code, metric names → **English**.\n- Default; follow the user's latest instruction if it signals otherwise.\n  Full contract: `skill-authoring-edho-ferdian` §7.\n\n## Workflow overview\n\n```\nPhase 0  Stack detection & baseline scope\nPhase 1  Baseline measurement            → references\u002Fweb-frontend.md (JS\u002FReact\u002FNext)\nPhase 2  Diagnosis against budgets       → references\u002Fweb-frontend.md §Core Web Vitals budget\n                                            references\u002Fweb-frontend.md §Algorithmic complexity\n                                            references\u002Fweb-frontend.md §Heap-snapshot diffing\nPhase 3  Fix (adaptive, preserve behavior)\nPhase 4  Re-measure & compute delta\nPhase 5  Report                          → references\u002Faudit-output-format.md\n```\n\n---\n\n## Phase 0 — Stack detection & baseline scope\n\nDetect the stack the same way the rest of this ecosystem does (manifest\nfiles) — this skill's ported content is JS\u002FReact\u002FNext-biased because that\nmatches Edho's actual stack (per `references\u002Fweb-frontend.md`); other stacks\nare listed as planned at the bottom of this file, not silently assumed to\nwork the same way.\n\nIdentify the target: a specific page\u002Froute, a specific function\u002Falgorithm, a\nbundle, memory behavior over time, or a database query — the target\ndetermines which subsection of Phase 1\u002F2 runs. Ask one question only if the\nuser's complaint (\"it's slow\") doesn't narrow this down at all — a vague\ncomplaint plus a quick look at the repo (routes, recent changes, an obvious\nhot path) usually narrows it without asking.\n\nProbe what's actually runnable: Lighthouse\u002FLighthouse CI, a bundle analyzer,\nChrome DevTools access (via the browser tools this environment provides),\na Node\u002Fbrowser profiler, `EXPLAIN ANALYZE` on a reachable database. Record\nwhat's available — this gates which Phase 1 measurements can produce\n**[High confidence]** results versus which have to be marked as needing a\ntool that isn't present.\n\n---\n\n## Phase 1 — Baseline measurement\n\n**Never skip straight to Phase 2's diagnosis without a number in hand.** The\nwhole reason this skill exists separately from the review skill's PERF\ndomain is that a number beats a guess. Run the tooling appropriate to the\ntarget identified in Phase 0:\n\n- **Page\u002Froute (Core Web Vitals)** — run Lighthouse (`npx lighthouse \u003Curl>\n  --view` or `--output=json` for a machine-readable baseline) and\u002For read\n  real-user Web Vitals if instrumented. Full detail in\n  `references\u002Fweb-frontend.md`.\n- **Bundle size** — `npx source-map-explorer` \u002F `npx webpack-bundle-analyzer`\n  against the actual production build, not dev mode.\n- **A specific function\u002Falgorithm** — profile it directly (Node\n  `--prof`\u002F`--prof-process`, or a simple timed harness for a pure function)\n  rather than reasoning about Big-O from reading it — confirm the actual\n  input size in production before concluding an O(n²) pattern matters at all.\n- **Memory leak suspicion** — two heap snapshots bracketing the suspected\n  leaking action; see `references\u002Fweb-frontend.md` §Heap-snapshot diffing\n  for the exact methodology (comparing snapshots is where most naive\n  attempts go wrong).\n- **Database query** — `EXPLAIN ANALYZE` against a reachable dev database;\n  cross-reference `database-lens.md` in `code-review-edho-ferdian` for the\n  static half of index\u002Fschema reasoning — this skill supplies the dynamic\n  half (actual query plan and timing).\n- **Backend\u002FAPI endpoint latency or throughput** — run the API benchmark mode\n  below (per-endpoint p50\u002Fp95\u002Fp99, serial then concurrent) or profile the\n  bulk-data-movement path; full detail in\n  `references\u002Fbackend-latency-and-throughput.md`.\n\nRecord the baseline number(s) precisely — this is what Phase 4's delta is\ncomputed against. If a needed tool genuinely isn't available, say so\nexplicitly and mark the resulting diagnosis at reduced confidence rather\nthan inventing a plausible-looking number.\n\n---\n\n## Phase 2 — Diagnosis against budgets and patterns\n\nWith a real baseline in hand, diagnose against:\n\n- **Core Web Vitals budget table** (with per-metric remediation steps) —\n  `references\u002Fweb-frontend.md` §Core Web Vitals budget.\n- **Algorithmic-complexity table** (bad pattern → better-complexity\n  replacement, e.g. repeated linear search in a loop → build a Map\u002FSet\n  first; nested loop over the same collection → single pass with an\n  accumulator) — `references\u002Fweb-frontend.md` §Algorithmic complexity.\n- **Bundle composition** — largest contributors, duplicate dependencies,\n  heavy libraries with lighter alternatives — `references\u002Fweb-frontend.md`\n  §Bundle optimization.\n- **Memory** — heap-snapshot diff results against the leak-pattern catalogue\n  (uncleaned listeners\u002Ftimers\u002Fsubscriptions, closures holding large refs) —\n  `references\u002Fweb-frontend.md` §Heap-snapshot diffing.\n\nEvery diagnosis cites the specific measurement that confirmed it (a\nLighthouse metric value, a query count, a heap delta in MB, a profiler\nsample) — a diagnosis without a cited number belongs back in the review\nskill's PERF domain as a static suspicion, not here as a confirmed finding.\n\n---\n\n## Phase 3 — Fix\n\nApply the fix. Same intent-preservation discipline as the rest of this\necosystem: fix the performance characteristic, don't change observable\nbehavior unless that's the explicit ask. Prefer the smallest change that\naddresses the measured bottleneck over a broad rewrite — a targeted `Map`\ninstead of a nested loop, a `useMemo` on the specific expensive computation,\nan added index, a `React.lazy` boundary on the specific heavy component —\nnot a speculative refactor of surrounding code that wasn't measured as slow.\n\n---\n\n## Optimization-loop pattern (try N variants, measure, pick the winner)\n\nFor a performance problem with more than one plausible fix and no obvious\nwinner (a query that could be indexed three different ways, a component\nthat could memoize at two different boundaries), it is often faster to try\na small number of variants and measure than to reason to a single answer\nup front:\n\n1. Generate 2-4 concrete variants of the fix.\n2. Measure each against the same benchmark\u002Fprofile, under the same\n   conditions (same input size, same machine state).\n3. Pick the winner by the actual measurement, not by which one looks more\n   elegant.\n4. Record what was tried and why the others lost — a future session\n   re-attempting an already-rejected variant wastes the same effort twice.\n\nThis is a small, bounded form of the loop pattern `gan-harness-edho-ferdian`\nuses more heavily — it does not need that skill's full generator\u002Fevaluator\nmachinery for a scoped performance decision, but the discipline of\n\"measure, don't guess, and record the losers\" is the same one.\n\n---\n\n### Mode — API benchmark\nHit each endpoint ~100 times, report p50\u002Fp95\u002Fp99 plus response size and status\ndistribution; then repeat at ~10 concurrent to expose contention that a serial\nrun hides. Compare against a stated SLA, not against \"feels fine\". Backend\nlatency\u002Fthroughput reasoning behind this mode lives in\n`references\u002Fbackend-latency-and-throughput.md`.\n\n### Mode — Development-loop benchmark\nCold build, hot-reload (HMR), full test-suite duration, type-check time, lint\ntime, container build time. Developer-loop latency is a performance budget\ntoo: a 40s HMR silently changes how the team works.\n\n### Baselines\nPersist baselines as JSON in the repo so they are shared and diffable, and\nreport before\u002Fafter as a delta table with an explicit verdict per row\n(BETTER \u002F WARN \u002F REGRESSION). A benchmark with no committed baseline can only\nproduce absolutes, never a regression signal.\n\n---\n\n## Phase 4 — Re-measure & compute delta\n\nRe-run the **same measurement** from Phase 1 against the same target, same\nmethodology (same Lighthouse flags, same profiler, same query, same action\nsequence for a heap-snapshot pair). Compute the delta and check it against\nthe budget from Phase 2. A fix that doesn't move the measured number, or\nmoves it but still misses the budget, is reported as such — don't round up\nto \"fixed\" on a partial improvement.\n\nIf the environment supports Lighthouse CI gating (a CI step that fails the\nbuild on a budget regression), note whether one exists and offer to wire the\nmeasured budget into it — see `references\u002Fweb-frontend.md` §Lighthouse CI\ngating for the approach.\n\n---\n\n## Phase 5 — Report\n\nFill in **`references\u002Faudit-output-format.md`**'s baseline → change → measured\ndelta → pass\u002Ffail structure and present it. Save it to the repo (e.g.\n`.\u002F\u003Ctarget>-performance-audit.md`) and tell the user the path, same pattern\nas `code-review-edho-ferdian`'s saved report.\n\n## Global rules\n\n1. **Measure before diagnosing. Measure after fixing.** No exceptions — a\n   number-free \"should be faster now\" is not a finding this skill produces.\n2. **Cite the tool and the number** for every diagnosis and every claimed\n   improvement.\n3. **Preserve intent.** Fix the measured characteristic, not the surrounding\n   design, unless asked.\n4. **Say when a tool isn't available** rather than inventing a plausible\n   number; mark confidence accordingly.\n5. **Language routing** as defined above.\n6. **Save the report file** at the end, same as the review skill.\n\n---\n\n## Additional references\n\nThis skill's ported content is JS\u002FReact\u002FNext-biased (matches Edho's current\nstack — `references\u002Fweb-frontend.md`). Backend\u002FAPI latency and throughput\n(endpoint p50\u002Fp95\u002Fp99, bulk data movement) is covered by\n`references\u002Fbackend-latency-and-throughput.md`.\n\n- `references\u002Fdatabase-query-profiling.md` — the dynamic half of database\n  performance work (cross-reference `database-lens.md` in\n  `code-review-edho-ferdian` for the static\u002Fschema half): reading\n  `EXPLAIN ANALYZE` output in depth, query-plan regression detection via\n  `pg_stat_statements`\u002Fplan-JSON diffing, slow-query-log analysis, connection-\n  pool saturation diagnosis, and runtime N+1 confirmation.\n- `references\u002Fnon-node-flamegraph-profiling.md` — flame-graph profiling for\n  non-Node backends: on-CPU vs off-CPU (wall-clock) profiling, reading a\n  flame graph (wide plateaus, tall towers), and tool-specific workflows for\n  Python (`py-spy`), Go (`pprof`), and Java (`async-profiler`).\n",{"name":5,"metadata":24,"description":7},{"last_reviewed":25},"2026-10-09","main","533d32ad55c92d40fbabe93cbe21fe048598fe5a",".zcode\u002Fskills\u002Fperformance-audit-edho-ferdian",0,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF","https:\u002F\u002Fgithub.com\u002Fedhoferdian","Elevating ideas through AI, rising from humble beginnings to mastering the frontiers of the web. Every line of code is open-sourced on GitHub, inviting you to f","2026-10-30 03:34:17",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add edhoferdian\u002FEEF --skill performance-audit-edho-ferdian",{"tool":40,"command":41},"bunx","bunx skills add edhoferdian\u002FEEF --skill performance-audit-edho-ferdian",{"tool":43,"command":44},"pnpm","pnpm dlx skills add edhoferdian\u002FEEF --skill performance-audit-edho-ferdian","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002FSKILL.md",true,[48,54,58,62,66,70],{"path":49,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":52,"sourceHtmlUrl":53},"references\u002Faudit-output-format.md","REFERENCE","参考文档",2829,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002Freferences\u002Faudit-output-format.md",{"path":55,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":56,"sourceHtmlUrl":57},"references\u002Fbackend-latency-and-throughput.md",8668,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002Freferences\u002Fbackend-latency-and-throughput.md",{"path":59,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":60,"sourceHtmlUrl":61},"references\u002Fdatabase-query-profiling.md",16031,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002Freferences\u002Fdatabase-query-profiling.md",{"path":63,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":64,"sourceHtmlUrl":65},"references\u002Fnon-node-flamegraph-profiling.md",13998,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002Freferences\u002Fnon-node-flamegraph-profiling.md",{"path":67,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":68,"sourceHtmlUrl":69},"references\u002Freact-nextjs.md",8938,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002Freferences\u002Freact-nextjs.md",{"path":71,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":72,"sourceHtmlUrl":73},"references\u002Fweb-frontend.md",10032,"https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002F.zcode\u002Fskills\u002Fperformance-audit-edho-ferdian\u002Freferences\u002Fweb-frontend.md",[75,82,89,96,103,110],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M7C9FY2EWE5NB7M3YW0KJX62","video-style-edho-ferdian","Measure a reference video's look and pacing (colour grade as a 3D LUT, cut rhythm, overlays) and reproduce it: grade, cut and composite new footage, verify numerically, hand off an NLE timeline. Use for \"tiru look video ini\", \"samakan grading\", \"bikin video dengan gaya referensi\", \"LUT dari referensi\", \"cut rhythm\".","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fvideo-style-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M7C9FT74P7JXGQ73116B9WS1","test-authoring-edho-ferdian","Write unit and component tests well: React Testing Library, pytest, Go and Vue, plus regression-test patterns. Use for \"tulis test untuk component ini\", \"bagaimana test hook ini\", \"test yang bagus untuk fitur X\", \"tulis test pytest\u002FGo\u002FVue\", or dev-kickoff's TEST stage.","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Ftest-authoring-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":19},"01M7C9FMPE9W76FS7S8ZZ48ARE","system-design-edho-ferdian","Mid-project architecture decisions: ADRs, trade-off analysis, non-functional requirements, scaling tiers and blast radius. Use for \"desain arsitektur\", \"keputusan teknis besar\", \"bikin ADR\", \"trade-off antara X dan Y\", \"microservices atau monolith\", \"apa yang rusak kalau X diubah\", or a capacity question.","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fsystem-design-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":19},"01M7C9FETATAHKG5WKPT4AP1A2","spec-mining-edho-ferdian","Extract behavioural specs from a codebase with none: WHEN\u002FTHEN requirements and invariants, each anchored to the code that enforces it, unclear behaviour recorded as uncertainty. Use for \"ekstrak spec\", \"buat spec dari kode\", \"reverse-engineer the spec\", \"repo ini tidak ada dokumentasi\", or a missing BEHAVIOR_SPEC in dev-kickoff.","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fspec-mining-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":19},"01M7C9F93YCXFCVRDG4S5QFAH4","skill-authoring-edho-ferdian","Create and govern this ecosystem's skills: search before building, a quality bar, compliance checks, promoting repeated principles to rules, and packaging dist\u002F*.skill. Use for \"bikin skill baru\", \"ada skill buat X gak\", \"fork skill ini\", \"skill gue kepake gak sih\", \"package skill ini\", \"buatkan .skill-nya\".","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fskill-authoring-edho-ferdian\u002FSKILL.md",[],[],{"skillNo":111,"name":112,"nameZh":6,"description":113,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":114,"categories":115,"keywords":116,"publishedAt":19},"01M7C9F4XZJ59R0GS4YEP4E1RD","skill-audit-edho-ferdian","Audit this repo's skills\u002F for staleness, redundancy, broken cross-references and weak descriptions (not the ~\u002F.claude setup: that is config-hygiene). Use for \"audit skill saya\", \"cek skill yang sudah dibuat\", \"ada yang redundan gak\", \"skill mana yang basi\".","https:\u002F\u002Fgithub.com\u002Fedhoferdian\u002FEEF\u002Fblob\u002Fmain\u002Fskills\u002Fskill-audit-edho-ferdian\u002FSKILL.md",[],[]]