[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M754G7VD9JBS4NVFNGREHP64":3,"skill-same:01M754G7VD9JBS4NVFNGREHP64":63},{"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":33,"commitSha":34,"skillDirectory":35,"forksCount":36,"repositoryHtmlUrl":37,"repositoryHomepageUrl":38,"ownerHtmlUrl":39,"ownerBlogUrl":40,"ownerDescription":41,"updatedAt":42,"install":43,"resources":56},"01M754G7VD9JBS4NVFNGREHP64","perf-optimization",null,"Profile-driven performance optimization of mecatl using the offline benchmark + scenario harness. Use when optimizing allocations or latency, reducing allocs\u002Fop or memory, profiling a Go benchmark, pinpointing a hotspot with pprof, proving a win with benchstat, adding a regression benchmark, wiring profile-guided optimization (PGO), or investigating \"why is this slow \u002F allocating\" or a suspected perf regression. Covers task bench, task perf:scenarios, memprofile -> pprof, benchstat A\u002FB, allocs-first gating, PGO setup, and the discipline (follow the profile not the hypothesis; keep pure-perf changes byte-identical; mutation-test cache guards; skip the wrong abstraction). NOT for the live perf MCP server (use perf-mcp-interpretation) or non-mecatl Go profiling.","stacklok\u002Fmecatl","stacklok","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F110237746?v=4","01M754F1F8WJJZYJ0WZ73C7PBS","Stacklok","01M754F1GGDR6FG8K1RZ8SR2NA",309,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fperf-optimization\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: perf-optimization\ndescription: >-\n  Profile-driven performance optimization of mecatl using the offline benchmark +\n  scenario harness. Use when optimizing allocations or latency, reducing allocs\u002Fop\n  or memory, profiling a Go benchmark, pinpointing a hotspot with pprof, proving a\n  win with benchstat, adding a regression benchmark, wiring profile-guided\n  optimization (PGO), or investigating \"why is this slow \u002F allocating\" or a suspected\n  perf regression. Covers task bench, task perf:scenarios, memprofile -> pprof,\n  benchstat A\u002FB, allocs-first gating, PGO setup, and the discipline (follow the\n  profile not the hypothesis; keep pure-perf changes byte-identical; mutation-test\n  cache guards; skip the wrong abstraction).\n  NOT for the live perf MCP server (use perf-mcp-interpretation) or non-mecatl Go\n  profiling.\nmetadata:\n  author: mecatl\n  tags: [performance, profiling, benchmarks, pprof, benchstat, allocations]\n---\n\n# Perf optimization (mecatl offline harness)\n\nThe companion to the regression-tracking design in\n[`docs\u002Fperf-tracking.md`](..\u002F..\u002F..\u002Fdocs\u002Fperf-tracking.md). This skill\nis the **offline benchmark\u002Fscenario** workflow: measure → profile → optimize →\nprove → guard. For diagnosing a **running** harness via the perf MCP server, use\nthe `perf-mcp-interpretation` skill instead — different tool, different signals.\n\n## The harness (where the numbers come from)\n\n- `task bench` — hot-path microbenchmarks (`engine\u002Fprompt`, `engine\u002Fgovernance`,\n  `engine\u002Fagent`). `BENCHCOUNT` default 10. Offline (mockllm + memfs). Not part of\n  `task test`. **The engine is its own Go module**, so `task bench`\u002F`task fuzz`\n  run these as `cd engine && go test … .\u002Fprompt\u002F .\u002Fgovernance\u002F .\u002Fagent\u002F`. An ad-hoc\n  re-run on an engine package must do the same: `cd engine && go test -bench=… .\u002Fagent\u002F`\n  (an explicit `.\u002Fengine\u002Fagent\u002F` path also resolves via the committed `go.work`, but the\n  `.\u002F...` wildcard does **not** cross the module boundary).\n- `task perf:scenarios` — five whole-loop scenarios. Four live in\n  `perf\u002Fscenarios\u002F` (single-session-long, team-fanout, background-subagents,\n  compaction-cycle); the fifth (tui-scrollback) lives in `cmd\u002Fmecatui\u002Fui\u002F`, and the\n  task runs **both** packages — so `go test .\u002Fperf\u002Fscenarios\u002F` alone gives only four.\n  `BENCHCOUNT` default 6. `MECATL_PERF_JSON=.scratch\u002Fx.json` writes the KPI JSON.\n- Both are deterministic and offline — never reach for a live model\u002Fnetwork.\n- **PGO\u002Fbench captures over `engine\u002F` packages depend on the active `go.work`** (the committed\n  workspace wires `.\u002Fengine` in alongside the root) — a `GOWORK=off` invocation resolves the engine\n  module standalone and won't see the host repo.\n\n## What to gate on (allocs-first)\n\n`allocs\u002Fop`, `bytes\u002Fop`, goroutine-delta, and cache-hit-rate are **deterministic and\nportable** — these are the real signal. `ns\u002Fop` and `rss_*` are machine-specific —\ntreat as advisory shape, never the headline. A perf change is **done only when\nallocs\u002Fop moves** in benchstat; a wall-clock-only \"win\" on a shared machine is noise.\n\n## Workflow\n\n### 1. Measure — pick the benchmark that actually exercises the path\n\nRun the relevant benchmark and confirm it reflects the workload you care about. A\nbenchmark whose per-iteration **setup** dwarfs the code under test, or that is\n**all-miss by construction**, will hide a real win. If no benchmark covers the path,\nadd one first (match the existing `bench_test.go` style: `for b.Loop()`, results to a\npackage-level sink, offline).\n\n### 2. Profile — pinpoint the site, do NOT guess\n\nCapture an allocation profile on the benchmark and follow it to a `file:line`:\n\n```sh\ngo test -run='^$' -bench=BenchmarkX -benchmem -memprofile=.scratch\u002Fx.mprof -count=3 .\u002Fpkg\u002F\ngo tool pprof -alloc_space   -top -nodecount=25 .scratch\u002Fx.mprof   # bytes\ngo tool pprof -alloc_objects -top -nodecount=25 .scratch\u002Fx.mprof   # object count\ngo tool pprof -list=FuncName .scratch\u002Fx.mprof                      # line-level\n```\n\nSave profiles under `.scratch\u002F` (repo rule — never `\u002Ftmp`). **Follow the profile to\nthe real site.** The hypothesis is often wrong (see the playbook: the TUI hotspot was\nthe string join, not `SetContent` as assumed). Let `-list` show you the exact lines.\n\n### 3. Design — the smallest change at the proven hotspot\n\nOptimize only what the profile proves is hot. Weigh the win against the real cost: a\nmicrosecond saved once per turn is invisible next to an LLM round-trip. **The wrong\nabstraction is worse than the allocation** — if the clean seam doesn't exist or the\nfix adds stateful invalidation surface for a marginal gain, it is a legitimate NO-GO.\nSay so and skip it rather than forcing it.\n\n### 4. Prove — benchstat before\u002Fafter, count >= 10\n\n```sh\ngo test -run='^$' -bench=BenchmarkX -benchmem -count=10 .\u002Fpkg\u002F > .scratch\u002Fbefore.txt\n# ... apply the change ...\ngo test -run='^$' -bench=BenchmarkX -benchmem -count=10 .\u002Fpkg\u002F > .scratch\u002Fafter.txt\nbenchstat .scratch\u002Fbefore.txt .scratch\u002Fafter.txt   # go install golang.org\u002Fx\u002Fperf\u002Fcmd\u002Fbenchstat@latest\n```\n\n`allocs\u002Fop` \u002F `B\u002Fop` must drop with a statistically significant delta. Re-run\n`task perf:scenarios` and confirm the scenario KPI moved in the expected direction.\nUpdate the baseline snapshot in `docs\u002Fperf-tracking.md`.\n\n### 5. Guard — keep behaviour identical, prove the guard isn't vacuous\n\n- **Pure-perf changes must be byte-identical.** For the TUI, `task test:golden` must\n  stay green with **zero golden diffs** — you change HOW, never WHAT. Caveat: golden\n  refresh runs `-update` first, so a plain `go test .\u002F...` (no `-update`) against the\n  committed goldens is what actually catches a regression; don't rely on the refresh\n  step to catch a stale serve.\n- **Any cache\u002Foracle you add must be mutation-tested.** Temporarily break it (force\n  the stale\u002Fwrong path), confirm a test goes red, then restore (back up with `cp`,\n  restore with `cp` — never `git checkout`, it wipes uncommitted work). A guard that\n  stays green when the behaviour is broken is worse than none.\n- **A perf change touching an EXPORTED `engine\u002F` symbol trips the `api-compat` CI gate**\n  (`task api:check`) — a failure mode a perf optimizer wouldn't expect. Keep\n  pure-perf changes byte-identical to the engine's public surface and it never fires; if the\n  surface legitimately changed, run `task api:update` and add an `engine\u002FCHANGELOG.md` entry\n  classified per `engine\u002FCOMPATIBILITY.md`.\n\n## Common pitfalls\n\n- **All-miss benchmark hides the win.** A streaming bench that mutates state every\n  iteration never hits the cache you added — it's the worst-case floor, flat by\n  design. Add a *steady-state* bench for the cache-hit path to show the real win.\n- **Setup swamps the signal.** Build fixtures outside `b.Loop()`; confirm with a\n  quick profile that setup isn't the dominant allocator.\n- **Chasing `ns\u002Fop` on a shared machine.** Gate on allocs; ns is advisory.\n- **Memoizing across the byte-stable prompt prefix.** Any prompt-inventory cache must\n  produce byte-identical output or it tanks the provider cache-hit rate — the exact\n  thing perf-tracking exists to protect. Usually not worth it (see playbook).\n\n## Complementary: PGO (free compiler-level wins)\n\nBeyond hand-optimizing a hot path, Profile-Guided Optimization lets the compiler\noptimize from a CPU profile (typically 2–14% CPU — but here sub-1% of wall-clock,\nsince cost is network-dominated: a free set-and-forget win, not a latency feature).\nThe mechanism is already wired:\n\n- `task pgo:collect` builds a PROVISIONAL offline profile under `.scratch\u002Fpgo\u002F`.\n- `cmd\u002Fmecated\u002Fdefault.pgo` is the reserved slot — `go build`'s `-pgo=auto` applies it\n  automatically the moment a profile is dropped there (nothing in the build chain\n  passes `-pgo=off`).\n- **Do NOT commit an offline-collected profile** — it trains the compiler on the\n  mockllm path that ships in no production binary (it can't pessimize, but it wastes\n  the one slot). Commit only a real `\u002Fdebug\u002Fpprof\u002Fprofile` capture from a running\n  `mecated` under load.\n\nFull rationale, the per-binary decision, and the production refresh + staleness\nprocess live in [`perf-tracking.md` \"Profile-guided optimization\"](..\u002F..\u002F..\u002Fdocs\u002Fperf-tracking.md#profile-guided-optimization) —\nread it before touching PGO.\n\n## See Also\n\n- [`references\u002Fplaybook.md`](references\u002Fplaybook.md) — pprof flag cookbook + two\n  worked examples (a real win and a real NO-GO) showing the discipline end to end.\n- [`docs\u002Fperf-tracking.md`](..\u002F..\u002F..\u002Fdocs\u002Fperf-tracking.md) — the KPI\n  design, gating posture, and baselines.\n- `perf-mcp-interpretation` skill — the **live** counterpart (running-harness\n  diagnosis via the perf MCP server).\n","# Perf optimization (mecatl offline harness)\n\nThe companion to the regression-tracking design in\n[`docs\u002Fperf-tracking.md`](..\u002F..\u002F..\u002Fdocs\u002Fperf-tracking.md). This skill\nis the **offline benchmark\u002Fscenario** workflow: measure → profile → optimize →\nprove → guard. For diagnosing a **running** harness via the perf MCP server, use\nthe `perf-mcp-interpretation` skill instead — different tool, different signals.\n\n## The harness (where the numbers come from)\n\n- `task bench` — hot-path microbenchmarks (`engine\u002Fprompt`, `engine\u002Fgovernance`,\n  `engine\u002Fagent`). `BENCHCOUNT` default 10. Offline (mockllm + memfs). Not part of\n  `task test`. **The engine is its own Go module**, so `task bench`\u002F`task fuzz`\n  run these as `cd engine && go test … .\u002Fprompt\u002F .\u002Fgovernance\u002F .\u002Fagent\u002F`. An ad-hoc\n  re-run on an engine package must do the same: `cd engine && go test -bench=… .\u002Fagent\u002F`\n  (an explicit `.\u002Fengine\u002Fagent\u002F` path also resolves via the committed `go.work`, but the\n  `.\u002F...` wildcard does **not** cross the module boundary).\n- `task perf:scenarios` — five whole-loop scenarios. Four live in\n  `perf\u002Fscenarios\u002F` (single-session-long, team-fanout, background-subagents,\n  compaction-cycle); the fifth (tui-scrollback) lives in `cmd\u002Fmecatui\u002Fui\u002F`, and the\n  task runs **both** packages — so `go test .\u002Fperf\u002Fscenarios\u002F` alone gives only four.\n  `BENCHCOUNT` default 6. `MECATL_PERF_JSON=.scratch\u002Fx.json` writes the KPI JSON.\n- Both are deterministic and offline — never reach for a live model\u002Fnetwork.\n- **PGO\u002Fbench captures over `engine\u002F` packages depend on the active `go.work`** (the committed\n  workspace wires `.\u002Fengine` in alongside the root) — a `GOWORK=off` invocation resolves the engine\n  module standalone and won't see the host repo.\n\n## What to gate on (allocs-first)\n\n`allocs\u002Fop`, `bytes\u002Fop`, goroutine-delta, and cache-hit-rate are **deterministic and\nportable** — these are the real signal. `ns\u002Fop` and `rss_*` are machine-specific —\ntreat as advisory shape, never the headline. A perf change is **done only when\nallocs\u002Fop moves** in benchstat; a wall-clock-only \"win\" on a shared machine is noise.\n\n## Workflow\n\n### 1. Measure — pick the benchmark that actually exercises the path\n\nRun the relevant benchmark and confirm it reflects the workload you care about. A\nbenchmark whose per-iteration **setup** dwarfs the code under test, or that is\n**all-miss by construction**, will hide a real win. If no benchmark covers the path,\nadd one first (match the existing `bench_test.go` style: `for b.Loop()`, results to a\npackage-level sink, offline).\n\n### 2. Profile — pinpoint the site, do NOT guess\n\nCapture an allocation profile on the benchmark and follow it to a `file:line`:\n\n```sh\ngo test -run='^$' -bench=BenchmarkX -benchmem -memprofile=.scratch\u002Fx.mprof -count=3 .\u002Fpkg\u002F\ngo tool pprof -alloc_space   -top -nodecount=25 .scratch\u002Fx.mprof   # bytes\ngo tool pprof -alloc_objects -top -nodecount=25 .scratch\u002Fx.mprof   # object count\ngo tool pprof -list=FuncName .scratch\u002Fx.mprof                      # line-level\n```\n\nSave profiles under `.scratch\u002F` (repo rule — never `\u002Ftmp`). **Follow the profile to\nthe real site.** The hypothesis is often wrong (see the playbook: the TUI hotspot was\nthe string join, not `SetContent` as assumed). Let `-list` show you the exact lines.\n\n### 3. Design — the smallest change at the proven hotspot\n\nOptimize only what the profile proves is hot. Weigh the win against the real cost: a\nmicrosecond saved once per turn is invisible next to an LLM round-trip. **The wrong\nabstraction is worse than the allocation** — if the clean seam doesn't exist or the\nfix adds stateful invalidation surface for a marginal gain, it is a legitimate NO-GO.\nSay so and skip it rather than forcing it.\n\n### 4. Prove — benchstat before\u002Fafter, count >= 10\n\n```sh\ngo test -run='^$' -bench=BenchmarkX -benchmem -count=10 .\u002Fpkg\u002F > .scratch\u002Fbefore.txt\n# ... apply the change ...\ngo test -run='^$' -bench=BenchmarkX -benchmem -count=10 .\u002Fpkg\u002F > .scratch\u002Fafter.txt\nbenchstat .scratch\u002Fbefore.txt .scratch\u002Fafter.txt   # go install golang.org\u002Fx\u002Fperf\u002Fcmd\u002Fbenchstat@latest\n```\n\n`allocs\u002Fop` \u002F `B\u002Fop` must drop with a statistically significant delta. Re-run\n`task perf:scenarios` and confirm the scenario KPI moved in the expected direction.\nUpdate the baseline snapshot in `docs\u002Fperf-tracking.md`.\n\n### 5. Guard — keep behaviour identical, prove the guard isn't vacuous\n\n- **Pure-perf changes must be byte-identical.** For the TUI, `task test:golden` must\n  stay green with **zero golden diffs** — you change HOW, never WHAT. Caveat: golden\n  refresh runs `-update` first, so a plain `go test .\u002F...` (no `-update`) against the\n  committed goldens is what actually catches a regression; don't rely on the refresh\n  step to catch a stale serve.\n- **Any cache\u002Foracle you add must be mutation-tested.** Temporarily break it (force\n  the stale\u002Fwrong path), confirm a test goes red, then restore (back up with `cp`,\n  restore with `cp` — never `git checkout`, it wipes uncommitted work). A guard that\n  stays green when the behaviour is broken is worse than none.\n- **A perf change touching an EXPORTED `engine\u002F` symbol trips the `api-compat` CI gate**\n  (`task api:check`) — a failure mode a perf optimizer wouldn't expect. Keep\n  pure-perf changes byte-identical to the engine's public surface and it never fires; if the\n  surface legitimately changed, run `task api:update` and add an `engine\u002FCHANGELOG.md` entry\n  classified per `engine\u002FCOMPATIBILITY.md`.\n\n## Common pitfalls\n\n- **All-miss benchmark hides the win.** A streaming bench that mutates state every\n  iteration never hits the cache you added — it's the worst-case floor, flat by\n  design. Add a *steady-state* bench for the cache-hit path to show the real win.\n- **Setup swamps the signal.** Build fixtures outside `b.Loop()`; confirm with a\n  quick profile that setup isn't the dominant allocator.\n- **Chasing `ns\u002Fop` on a shared machine.** Gate on allocs; ns is advisory.\n- **Memoizing across the byte-stable prompt prefix.** Any prompt-inventory cache must\n  produce byte-identical output or it tanks the provider cache-hit rate — the exact\n  thing perf-tracking exists to protect. Usually not worth it (see playbook).\n\n## Complementary: PGO (free compiler-level wins)\n\nBeyond hand-optimizing a hot path, Profile-Guided Optimization lets the compiler\noptimize from a CPU profile (typically 2–14% CPU — but here sub-1% of wall-clock,\nsince cost is network-dominated: a free set-and-forget win, not a latency feature).\nThe mechanism is already wired:\n\n- `task pgo:collect` builds a PROVISIONAL offline profile under `.scratch\u002Fpgo\u002F`.\n- `cmd\u002Fmecated\u002Fdefault.pgo` is the reserved slot — `go build`'s `-pgo=auto` applies it\n  automatically the moment a profile is dropped there (nothing in the build chain\n  passes `-pgo=off`).\n- **Do NOT commit an offline-collected profile** — it trains the compiler on the\n  mockllm path that ships in no production binary (it can't pessimize, but it wastes\n  the one slot). Commit only a real `\u002Fdebug\u002Fpprof\u002Fprofile` capture from a running\n  `mecated` under load.\n\nFull rationale, the per-binary decision, and the production refresh + staleness\nprocess live in [`perf-tracking.md` \"Profile-guided optimization\"](..\u002F..\u002F..\u002Fdocs\u002Fperf-tracking.md#profile-guided-optimization) —\nread it before touching PGO.\n\n## See Also\n\n- [`references\u002Fplaybook.md`](references\u002Fplaybook.md) — pprof flag cookbook + two\n  worked examples (a real win and a real NO-GO) showing the discipline end to end.\n- [`docs\u002Fperf-tracking.md`](..\u002F..\u002F..\u002Fdocs\u002Fperf-tracking.md) — the KPI\n  design, gating posture, and baselines.\n- `perf-mcp-interpretation` skill — the **live** counterpart (running-harness\n  diagnosis via the perf MCP server).\n",{"name":5,"metadata":24,"description":7},{"tags":25,"author":32},[26,27,28,29,30,31],"performance","profiling","benchmarks","pprof","benchstat","allocations","mecatl","main","c4ff786a7467eb2315de679ae78711d118775383",".claude\u002Fskills\u002Fperf-optimization",29,"https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl","https:\u002F\u002Fmecatl.dev","https:\u002F\u002Fgithub.com\u002Fstacklok","https:\u002F\u002Fstacklok.com","Open Source AI Innovation, Grounded in Trust.","2026-11-15 04:19:54",{"source":8,"skillName":5,"commands":44,"prompt":54,"zipAvailable":55},[45,48,51],{"tool":46,"command":47},"npx","npx skills add stacklok\u002Fmecatl --skill perf-optimization",{"tool":49,"command":50},"bunx","bunx skills add stacklok\u002Fmecatl --skill perf-optimization",{"tool":52,"command":53},"pnpm","pnpm dlx skills add stacklok\u002Fmecatl --skill perf-optimization","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fperf-optimization\u002FSKILL.md",true,[57],{"path":58,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":61,"sourceHtmlUrl":62},"references\u002Fplaybook.md","REFERENCE","参考文档",5338,"https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fperf-optimization\u002Freferences\u002Fplaybook.md",[64,71,78,85,92,99],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":19},"01M754GX1DCZR950M0SJFD048A","repo-fact","Use when asked for this repository's registered fact or codeword. Loads the canonical answer format.","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002Fe2e\u002Ffixtures\u002Fskills\u002Frepo-fact\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":19},"01M754GSSV3ACG56M3XWT5F2PD","greet","Use when asked to greet someone with the ceremonial mecatl greeting. Loads the exact greeting format.","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002Fe2e\u002Ffixtures\u002Fskills\u002Fgreet\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":19},"01M754GPQN80V07DGCPCB7P2XX","go-table-tests","Write idiomatic table-driven Go tests that stay offline and follow this repo's conventions.","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002Fdocs\u002Fexamples\u002Fskills\u002Fgo-table-tests\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":19},"01M754GMQM60ARRZ0251WD6JN8","commit-style","Write Conventional-Commits messages for this repo, with the required Co-Authored-By trailer.","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002Fdocs\u002Fexamples\u002Fskills\u002Fcommit-style\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":19},"01M754GGFCQ3AWR3HHHSN3C91C","tech-writer","Use when writing or substantively editing user-facing documentation: drafting a new page, rewriting or restructuring an existing one, adding a major section, or turning engineering material (PR descriptions, specs, release notes, rough notes) into docs. Writes clear, focused technical documentation following the Diataxis framework and Mecatl's canonical style guide. Use it even when the request doesn't mention writing quality; it governs how documentation gets written. Not for editorial review of finished work.","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Ftech-writer\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":19},"01M754G1M7ME416A88V3EX42GD","perf-mcp-interpretation","Interpret the mecatl perf MCP server's output to diagnose latency, goroutine leaks, GC pressure, allocation churn, and memory growth in a running mecatl harness. Use when connected to the mecatl perf MCP server (the perf:\u002F\u002F resources or the query_metric \u002F top_cpu_functions \u002F capture_cpu_profile \u002F top_allocations \u002F list_slow_turns \u002F capture_flight_recorder tools) and investigating why mecatl is slow, leaking, or growing. Covers tool routing\u002Fcost, reading pprof rankings and runtime metrics, and the leak\u002Fcontention\u002FGC signatures. NOT for generic Go profiling or non-mecatl MCP servers.","https:\u002F\u002Fgithub.com\u002Fstacklok\u002Fmecatl\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fperf-mcp-interpretation\u002FSKILL.md",[],[]]