[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7B2YNTJKTKQ80MM4TGRZN8K":3,"skill-same:01M7B2YNTJKTKQ80MM4TGRZN8K":70},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7B2YNTJKTKQ80MM4TGRZN8K","hamster-cpu-profiling",null,"Investigate main-thread freezes \u002F slow pages with a real CPU profile instead of guessing. Use when a page locks up, the sidebar\u002Fcards go unresponsive, requests hang pending, a 'freeze' or 'jank' is reported, or you need to prove a perf fix worked (before\u002Fafter). Drives a real browser via agent-browser, captures a Chrome DevTools profile, and parses it for the dominant hot frame — separating real app cost from profiler\u002Fdev-mode artifacts.","gethamster\u002Fskills","gethamster","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F215697643?v=4","01M7B2R6PGB3SDZZ71E0WE2S74","Hamster","01M7B2R6Q21HWJRXZZWB78BNDG",4,"MIT","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: hamster-cpu-profiling\ndescription: \"Investigate main-thread freezes \u002F slow pages with a real CPU profile instead of guessing. Use when a page locks up, the sidebar\u002Fcards go unresponsive, requests hang pending, a 'freeze' or 'jank' is reported, or you need to prove a perf fix worked (before\u002Fafter). Drives a real browser via agent-browser, captures a Chrome DevTools profile, and parses it for the dominant hot frame — separating real app cost from profiler\u002Fdev-mode artifacts.\"\nversion: \"1.0.0\"\n---\n\n# Hamster CPU profiling — find the freeze before theorizing\n\nA page freeze is a **main-thread CPU block**: one JavaScript task runs so long that paint, input, and pending fetches all stall behind it. You cannot fix it by reading code and guessing — the cost is rarely where it looks. **Get a real CPU profile first**, find the dominant frame, fix that, then re-profile to prove it.\n\n> First, classify the failure mode (see `feedback_verify_failure_mode_before_fixing`): a **CPU hang** (this skill) is different from a thrown error (error boundary) or a network stall (waterfall). The tell for a CPU hang: an in-page probe can't run during the block — you must measure from **outside** the page.\n\n## Tooling: agent-browser, NOT Playwright\n\nUse the `agent-browser` CLI. It drives a persistent, auth-preserving browser session and exposes `profiler start\u002Fstop`. Playwright's per-test isolation makes warm-cache \u002F post-deploy repro awkward, and you can't easily time round-trips from outside.\n\n```bash\nagent-browser --session-name \u003Cname> open \"\u003Curl>\"     # navigate (auth persists per session)\nagent-browser --session-name \u003Cname> profiler start    # begin CPU profile (DETACHES the page)\nagent-browser --session-name \u003Cname> profiler stop out.json\nagent-browser --session-name \u003Cname> eval \"1+1\"        # round-trip probe\n```\n\n## The method\n\n### 1. Open the page and confirm it rendered\n\n```bash\nagent-browser --session-name perf open \"http:\u002F\u002Flocalhost:8080\u002Fhome\u002F\u003Caccount>\u002F\u003Cpage>\"\nagent-browser --session-name perf eval \"document.querySelectorAll('\u003Crows-selector>').length\"\n```\n\nSession auth can expire — if you land on `\u002Fauth\u002Fsign-in`, fill `input[type=email]` \u002F `input[type=password]` and click the submit button (snapshot `-i -c` to get its ref; refs are NOT stable across reloads, re-snapshot each time). Always go through the proxy port (8080), never raw Next (3000).\n\n### 2. Detect the block from OUTSIDE (eval round-trip)\n\nA CPU hang can't run an in-page timer, so measure the wall-clock of an `eval` round-trip from the shell:\n\n```bash\nfor i in $(seq 1 20); do\n  t0=$(node -e 'process.stdout.write(String(Date.now()))')\n  agent-browser --session-name perf eval \"1+1\" >\u002Fdev\u002Fnull 2>&1\n  t1=$(node -e 'process.stdout.write(String(Date.now()))')\n  echo \"[$i] $((t1 - t0)) ms\"\ndone\n```\n\nBaseline is ~160–200 ms. A block shows as seconds. (A reload-triggered first eval also includes navigation time — don't read that one as pure CPU; use the trace's task durations instead, step 5.)\n\n### 3. Force the worst case (for local-first \u002F stash pages)\n\nLocal-first pages (`@kit\u002Fstash`) resolve from IndexedDB + react-query. The expensive path is a **cold resolve**: the id-list is on disk but the entity rows are not (a post-deploy cache bust, or a large cold account). Reproduce it by deleting entity blobs while keeping the id-list, then reload. Stash realtime tiers have a 30s `staleTime`, so wait ~35s after load before reloading so the list actually refetches. See `scripts\u002Fbust-stash-entities.js` for the IndexedDB cursor.\n\n> ⚠️ **Local dev only.** That helper *deletes* IndexedDB data — it's a repro tool for `localhost`, and self-aborts on any non-localhost origin. Never run it against a deployed environment.\n\n### 4. Capture the profile (mind the detach + reload caveats)\n\n`profiler start` **detaches** the page, so fire the reload AFTER starting, and **background it** — a foreground reload hangs ~2 min because analytics requests never let the page reach `networkidle`.\n\n```bash\nagent-browser --session-name perf profiler start\nagent-browser --session-name perf reload >\u002Fdev\u002Fnull 2>&1 &   # background\nRPID=$!\nsleep 9                                                        # capture window\nagent-browser --session-name perf profiler stop trace.json\nkill $RPID 2>\u002Fdev\u002Fnull\n```\n\n### 5. Parse — and separate signal from artifact\n\nThe bundled scripts in `scripts\u002F` aggregate the trace. **The cardinal rule: agent-browser's own instrumentation and React dev-mode logging are NOT your app's cost.** Strip them or you'll chase ghosts:\n\n- `(idle)` \u002F `(program)` \u002F `agentation*` → agent-browser overhead. It also **hooks `setTimeout`\u002F`clearTimeout`**, so native timer self-time is wildly inflated (~80µs\u002Fcall vs ~1µs in prod). Use timer *call counts* as a signal, never their µs.\n- `logComponentRender` \u002F `logComponentEffect` \u002F `measure` \u002F `run` → React **dev-mode** instrumentation, absent in production.\n- What's left — your app + library JS — is the real prod-relevant cost.\n\n```bash\nnode scripts\u002Flongtasks.cjs trace.json        # GROUND TRUTH: longest main-thread tasks (a >50ms task janks; >200ms is a visible freeze)\nnode scripts\u002Fclassify.cjs trace.json         # self-time split into app-JS vs artifacts; top prod-relevant frames\nnode scripts\u002Fcallers.cjs trace.json setTimeout   # who calls a hot native fn (catches agentation inflation)\nnode scripts\u002Fprofile-selftime.cjs trace.json     # raw self-time by function (no classification)\n```\n\nThe **longest-task** number from `longtasks.cjs` is the user-facing truth and the metric to quote before\u002Fafter. `classify.cjs` tells you *which frame* to attack.\n\n### 6. Fix the dominant frame, then PROVE it\n\nRe-run the exact same capture after the fix and compare longest-task + tasks>200ms + the top app-JS frames. Do not claim a fix without an after-profile showing the hot frame gone. A real example (briefs list, ~1000 items): longest task **620ms → 108ms**, tasks >200ms **2 → 0**, after replacing an N-observer `useQueries` with one cache subscription — the query-core observer frames vanished from the profile entirely.\n\n## Common traps\n\n- **Trusting raw self-time.** Native `clearTimeout` topped one profile at 196ms — but `callers.cjs` showed it was agent-browser's hook. Classify first.\n- **Per-render `JSON.stringify`\u002F`join` look guilty but profile at ~0ms** — V8 is fast; the real cost was observer machinery + re-render count. Measure, don't assume.\n- **Reading the post-reload first eval as CPU.** It includes navigation; use trace task durations.\n- **Dev vs prod skew.** `jsxDEV` and React dev logging inflate render cost in dev. The *structure* of the cost (which subsystem, how many re-renders) transfers to prod; the absolute µs does not. For a prod-accurate number, profile a production build.\n\n## Scripts\n\nAll in `scripts\u002F` (plain Node, no deps). Handle both `.cpuprofile` and DevTools-trace (`Profile`\u002F`ProfileChunk`) formats.\n\n| Script | Answers |\n| --- | --- |\n| `longtasks.cjs` | How bad is the block? (longest tasks, busy time, task count) |\n| `classify.cjs` | Where is the *real* cost? (app-JS vs artifact buckets, top frames) |\n| `callers.cjs \u003Cfn>` | Who calls this hot function? (de-attributes inflated native frames) |\n| `profile-selftime.cjs` | Raw self-time leaderboard (unclassified) |\n| `bust-stash-entities.js` | Force the cold-resolve worst case (eval body) |\n","# Hamster CPU profiling — find the freeze before theorizing\n\nA page freeze is a **main-thread CPU block**: one JavaScript task runs so long that paint, input, and pending fetches all stall behind it. You cannot fix it by reading code and guessing — the cost is rarely where it looks. **Get a real CPU profile first**, find the dominant frame, fix that, then re-profile to prove it.\n\n> First, classify the failure mode (see `feedback_verify_failure_mode_before_fixing`): a **CPU hang** (this skill) is different from a thrown error (error boundary) or a network stall (waterfall). The tell for a CPU hang: an in-page probe can't run during the block — you must measure from **outside** the page.\n\n## Tooling: agent-browser, NOT Playwright\n\nUse the `agent-browser` CLI. It drives a persistent, auth-preserving browser session and exposes `profiler start\u002Fstop`. Playwright's per-test isolation makes warm-cache \u002F post-deploy repro awkward, and you can't easily time round-trips from outside.\n\n```bash\nagent-browser --session-name \u003Cname> open \"\u003Curl>\"     # navigate (auth persists per session)\nagent-browser --session-name \u003Cname> profiler start    # begin CPU profile (DETACHES the page)\nagent-browser --session-name \u003Cname> profiler stop out.json\nagent-browser --session-name \u003Cname> eval \"1+1\"        # round-trip probe\n```\n\n## The method\n\n### 1. Open the page and confirm it rendered\n\n```bash\nagent-browser --session-name perf open \"http:\u002F\u002Flocalhost:8080\u002Fhome\u002F\u003Caccount>\u002F\u003Cpage>\"\nagent-browser --session-name perf eval \"document.querySelectorAll('\u003Crows-selector>').length\"\n```\n\nSession auth can expire — if you land on `\u002Fauth\u002Fsign-in`, fill `input[type=email]` \u002F `input[type=password]` and click the submit button (snapshot `-i -c` to get its ref; refs are NOT stable across reloads, re-snapshot each time). Always go through the proxy port (8080), never raw Next (3000).\n\n### 2. Detect the block from OUTSIDE (eval round-trip)\n\nA CPU hang can't run an in-page timer, so measure the wall-clock of an `eval` round-trip from the shell:\n\n```bash\nfor i in $(seq 1 20); do\n  t0=$(node -e 'process.stdout.write(String(Date.now()))')\n  agent-browser --session-name perf eval \"1+1\" >\u002Fdev\u002Fnull 2>&1\n  t1=$(node -e 'process.stdout.write(String(Date.now()))')\n  echo \"[$i] $((t1 - t0)) ms\"\ndone\n```\n\nBaseline is ~160–200 ms. A block shows as seconds. (A reload-triggered first eval also includes navigation time — don't read that one as pure CPU; use the trace's task durations instead, step 5.)\n\n### 3. Force the worst case (for local-first \u002F stash pages)\n\nLocal-first pages (`@kit\u002Fstash`) resolve from IndexedDB + react-query. The expensive path is a **cold resolve**: the id-list is on disk but the entity rows are not (a post-deploy cache bust, or a large cold account). Reproduce it by deleting entity blobs while keeping the id-list, then reload. Stash realtime tiers have a 30s `staleTime`, so wait ~35s after load before reloading so the list actually refetches. See `scripts\u002Fbust-stash-entities.js` for the IndexedDB cursor.\n\n> ⚠️ **Local dev only.** That helper *deletes* IndexedDB data — it's a repro tool for `localhost`, and self-aborts on any non-localhost origin. Never run it against a deployed environment.\n\n### 4. Capture the profile (mind the detach + reload caveats)\n\n`profiler start` **detaches** the page, so fire the reload AFTER starting, and **background it** — a foreground reload hangs ~2 min because analytics requests never let the page reach `networkidle`.\n\n```bash\nagent-browser --session-name perf profiler start\nagent-browser --session-name perf reload >\u002Fdev\u002Fnull 2>&1 &   # background\nRPID=$!\nsleep 9                                                        # capture window\nagent-browser --session-name perf profiler stop trace.json\nkill $RPID 2>\u002Fdev\u002Fnull\n```\n\n### 5. Parse — and separate signal from artifact\n\nThe bundled scripts in `scripts\u002F` aggregate the trace. **The cardinal rule: agent-browser's own instrumentation and React dev-mode logging are NOT your app's cost.** Strip them or you'll chase ghosts:\n\n- `(idle)` \u002F `(program)` \u002F `agentation*` → agent-browser overhead. It also **hooks `setTimeout`\u002F`clearTimeout`**, so native timer self-time is wildly inflated (~80µs\u002Fcall vs ~1µs in prod). Use timer *call counts* as a signal, never their µs.\n- `logComponentRender` \u002F `logComponentEffect` \u002F `measure` \u002F `run` → React **dev-mode** instrumentation, absent in production.\n- What's left — your app + library JS — is the real prod-relevant cost.\n\n```bash\nnode scripts\u002Flongtasks.cjs trace.json        # GROUND TRUTH: longest main-thread tasks (a >50ms task janks; >200ms is a visible freeze)\nnode scripts\u002Fclassify.cjs trace.json         # self-time split into app-JS vs artifacts; top prod-relevant frames\nnode scripts\u002Fcallers.cjs trace.json setTimeout   # who calls a hot native fn (catches agentation inflation)\nnode scripts\u002Fprofile-selftime.cjs trace.json     # raw self-time by function (no classification)\n```\n\nThe **longest-task** number from `longtasks.cjs` is the user-facing truth and the metric to quote before\u002Fafter. `classify.cjs` tells you *which frame* to attack.\n\n### 6. Fix the dominant frame, then PROVE it\n\nRe-run the exact same capture after the fix and compare longest-task + tasks>200ms + the top app-JS frames. Do not claim a fix without an after-profile showing the hot frame gone. A real example (briefs list, ~1000 items): longest task **620ms → 108ms**, tasks >200ms **2 → 0**, after replacing an N-observer `useQueries` with one cache subscription — the query-core observer frames vanished from the profile entirely.\n\n## Common traps\n\n- **Trusting raw self-time.** Native `clearTimeout` topped one profile at 196ms — but `callers.cjs` showed it was agent-browser's hook. Classify first.\n- **Per-render `JSON.stringify`\u002F`join` look guilty but profile at ~0ms** — V8 is fast; the real cost was observer machinery + re-render count. Measure, don't assume.\n- **Reading the post-reload first eval as CPU.** It includes navigation; use trace task durations.\n- **Dev vs prod skew.** `jsxDEV` and React dev logging inflate render cost in dev. The *structure* of the cost (which subsystem, how many re-renders) transfers to prod; the absolute µs does not. For a prod-accurate number, profile a production build.\n\n## Scripts\n\nAll in `scripts\u002F` (plain Node, no deps). Handle both `.cpuprofile` and DevTools-trace (`Profile`\u002F`ProfileChunk`) formats.\n\n| Script | Answers |\n| --- | --- |\n| `longtasks.cjs` | How bad is the block? (longest tasks, busy time, task count) |\n| `classify.cjs` | Where is the *real* cost? (app-JS vs artifact buckets, top frames) |\n| `callers.cjs \u003Cfn>` | Who calls this hot function? (de-attributes inflated native frames) |\n| `profile-selftime.cjs` | Raw self-time leaderboard (unclassified) |\n| `bust-stash-entities.js` | Force the cold-resolve worst case (eval body) |\n",{"name":5,"version":24,"description":7},"1.0.0","main","fe3bb1476c3ec23d8a366eac0fa8cd9f0852360a","skills\u002F.experimental\u002Fhamster-cpu-profiling",0,"https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills","https:\u002F\u002Fgithub.com\u002Fgethamster","https:\u002F\u002Ftryhamster.com","We're building the AI-native future of work, unlocking the promised acceleration for your whole team, from idea to impact.","2026-11-05 11:33:33",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add gethamster\u002Fskills --skill hamster-cpu-profiling",{"tool":40,"command":41},"bunx","bunx skills add gethamster\u002Fskills --skill hamster-cpu-profiling",{"tool":43,"command":44},"pnpm","pnpm dlx skills add gethamster\u002Fskills --skill hamster-cpu-profiling","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002FSKILL.md",true,[48,54,58,62,66],{"path":49,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":52,"sourceHtmlUrl":53},"scripts\u002Fbust-stash-entities.js","SCRIPT","脚本",2605,"https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002Fscripts\u002Fbust-stash-entities.js",{"path":55,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":56,"sourceHtmlUrl":57},"scripts\u002Fcallers.cjs",2282,"https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002Fscripts\u002Fcallers.cjs",{"path":59,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":60,"sourceHtmlUrl":61},"scripts\u002Fclassify.cjs",2900,"https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002Fscripts\u002Fclassify.cjs",{"path":63,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":64,"sourceHtmlUrl":65},"scripts\u002Flongtasks.cjs",1242,"https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002Fscripts\u002Flongtasks.cjs",{"path":67,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":68,"sourceHtmlUrl":69},"scripts\u002Fprofile-selftime.cjs",3213,"https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002F.experimental\u002Fhamster-cpu-profiling\u002Fscripts\u002Fprofile-selftime.cjs",[71,78,85,92,99,106],{"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},"01M7B5MWYW0KE251276S56KQDZ","writing-waterfall-requirements-documents","Write a waterfall requirements document: elicit, structure and review testable, traceable requirements, then baseline the specification before design.","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-waterfall-requirements-documents\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},"01M7B5MRCXWWPKREPB1CHQ7H90","writing-vision-and-values-statements","Write a V2MOM vision a reader can test and a ranked list of values that settles real tradeoffs, so the rest of the document has a clear anchor.","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-vision-and-values-statements\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},"01M7B5MKH2CH22YZAHQW3080VQ","writing-system-prompts-for-claude","Write Claude prompts and CLAUDE.md files that supply exact inputs, scope the task, and define done so the first attempt can be checked.","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-system-prompts-for-claude\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},"01M7B5ME1Y3X9RNRKP5HKHKACZ","writing-page-specific-website-copy","Website copywriting by page type: write homepage, landing page, pricing page and feature page copy that fits each page's visitor and single job.","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-page-specific-website-copy\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},"01M7B5M8MQNM8MN8PYH6BVWJ2B","writing-internal-press-releases","How to write an internal press release: the one-page working backwards press release that opens a PR\u002FFAQ document before anything is built.","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-internal-press-releases\u002FSKILL.md",[],[],{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":19},"01M7B5M3P3YH1XF37R8PWXXSBK","writing-email-sequences-that-sell","Email sequence copywriting for nurture and sales: give each email one job, write clear subject lines, and close with one specific call to action.","https:\u002F\u002Fgithub.com\u002Fgethamster\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-email-sequences-that-sell\u002FSKILL.md",[],[]]