[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6MF3ZR16S445SCS959E7GPB":3,"skill-same:01M6MF3ZR16S445SCS959E7GPB":72},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":22,"publishedAt":42,"linkOnly":43,"markdownContent":44,"markdownBody":45,"frontMatter":46,"refName":52,"commitSha":53,"skillDirectory":54,"forksCount":13,"repositoryHtmlUrl":55,"repositoryHomepageUrl":6,"ownerHtmlUrl":56,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":57,"install":58,"resources":71},"01M6MF3ZR16S445SCS959E7GPB","agent-memory",null,"Use when a project's CLAUDE.md has grown past what anyone reads and you want the agent to learn durable facts from its own sessions instead — or when asking why the agent keeps re-learning the same correction, why a remembered rule is wrong, or where a memory line came from. Implements a four-tier store (L0 transcripts \u002F L1 candidates \u002F L2 project context \u002F L3 stable persona) where promotion is earned by recurrence across sessions and days, never by one confident statement, and nothing reaches a committed file without a human adopting it.","JiayiV\u002Fquestion_mixing","JiayiV","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F54466889?v=4","01M6JFJYYMRZ3QRGP9339JNQ2K","01M6JFJZ35KB0QVT50NAHPJA7G",0,"MIT","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fpublic-02-claude-skills-main-gemini-skills-agent-memory-skill-e67d5358\u002FSKILL.md",[17],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[23,24,25,26,27,28,29,30,31,32,33,5,34,35,36,37,38,39,40,41],"agent memory","memory","hooks","prices","re","you","tencent","memory system","SkillOpt","project","claim","never","claude code hooks","improve","wiki","persona","database","code","learn","2026-11-10 10:12:30",false,"---\nname: agent-memory\ndescription: Use when a project's CLAUDE.md has grown past what anyone reads and you want the agent to learn durable facts from its own sessions instead — or when asking why the agent keeps re-learning the same correction, why a remembered rule is wrong, or where a memory line came from. Implements a four-tier store (L0 transcripts \u002F L1 candidates \u002F L2 project context \u002F L3 stable persona) where promotion is earned by recurrence across sessions and days, never by one confident statement, and nothing reaches a committed file without a human adopting it.\nargument-hint: \"[optional: status | why \\\"\u003Cclaim>\\\" | a tier name]\"\nlicense: MIT\nmetadata:\n  version: 1.0.0\n  build_pattern: \"Tencent TencentDB-Agent-Memory's tiering concept rebuilt natively on Claude Code hooks; deterministic recurrence gates, no LLM, no database\"\n  distinct_from: \"llm-wiki (a vault you write on purpose; this writes itself from sessions); skillopt-sleep (replays tasks to improve a skill; this extracts facts to remember); memory-engineering (audits and prices any memory system; this IS one, and is a legitimate subject of that audit)\"\n---\n\n# Agent Memory — promotion is earned, not asserted\n\n> **Portability:** stdlib only. No database, no embeddings, no network, no LLM calls.\n\n## The problem\n\nA project's `CLAUDE.md` is a memory system with one tier and no eviction: every\ndurable fact and every passing preference land in the same always-loaded file,\nuntil the important lines are diluted by the incidental ones. Facts learned\nmid-session vanish at teardown unless someone writes them down.\n\n**The fix is not more storage — it is a promotion ladder.** A claim earns its\nway toward always-loaded context by recurring; a human confirms the last step.\n\n## The four tiers\n\nTiers are distinguished by **injection policy**, not storage format.\n\n| Tier | Holds | Injected | Committed |\n|---|---|---|---|\n| **L0** | raw session transcripts | never | no (already on disk) |\n| **L1** | candidate atoms | on relevance, at prompt time | no (gitignored) |\n| **L2** | this project's context | every session start | yes, after adopt |\n| **L3** | stable cross-project persona | always | yes, after adopt |\n\n## The gates\n\nNothing moves up because it sounded important. It moves up because it recurred.\n\n- **L0 → L1** — an explicit marker fires (a directive, a correction, a stated\n  preference, a named lesson, a reproducible failure). Rule-based, high\n  precision, deliberately low recall.\n- **L1 → L2** — ≥ 3 distinct sessions spanning ≥ 2 distinct calendar days. A\n  claim stated outright needs 2 sessions; the distinct-day rule still applies. A\n  verified claim promotes on one observation and is the only day-exempt path.\n- **L2 → L3** — held in ≥ 2 distinct projects, aged ≥ 30 days, uncontested.\n\n**Two gates refuse rather than guess.** A claim whose text was altered by\nredaction never promotes on evidence alone — the flag firing is evidence the\nsource was sensitive, and a lexical filter finding one secret is not proof it\nfound all of them. A claim with an open contradiction is frozen at L1 until a\nhuman resolves it; the incumbent is never silently overwritten.\n\n## Use it\n\n```bash\n# what is remembered, and what is blocking the next promotion\npython3 scripts\u002Fmemory_inspect.py --tier L1\n\n# where did this line come from — sessions, days, transcript, quoted source\npython3 scripts\u002Fmemory_inspect.py --why \"PR base branch is dev\"\n\n# every claim with an open contradiction, both directions of the join\npython3 scripts\u002Fmemory_inspect.py --contested\n\n# dry-run the promotion pass; writes nothing\npython3 scripts\u002Fmemory_promote.py\n```\n\nThree hooks run the loop unattended: `SessionStart` injects L2 + L3,\n`UserPromptSubmit` recalls relevant L1 atoms, `SessionEnd` captures and stages.\nEach is disabled independently with `AGENT_MEMORY_SESSIONSTART=0`,\n`AGENT_MEMORY_USERPROMPTSUBMIT=0`, `AGENT_MEMORY_SESSIONEND=0`. Every hook fails\nopen: a broken memory system costs you memory, never a session.\n\n## Hard rules\n\n1. **Redact before writing.** Every atom passes the filter before it reaches\n   disk. Anything altered is quarantined from promotion.\n2. **Propose, never apply.** Promotions land in `.memory\u002Fstaged\u002F`. Only an\n   explicit `\u002Fcs:memory adopt` touches a `CLAUDE.md`, and it backs both up first.\n3. **Cite, don't invent.** Every atom carries a back-pointer to the transcript\n   line that produced it. `--why` resolving to *ambiguous* prints nothing rather\n   than guess: a wrong citation is worse than a missing one.\n4. **Never surface a contested claim as fact.** It is still injected — hiding\n   the conflict is worse — but always tagged.\n5. **The committed tiers carry no paths.** Promotion strips the back-pointer\n   prefix, which embeds an OS username.\n\n## Forcing questions\n\nWalk these one at a time before trusting the store.\n\n1. Which line in your `CLAUDE.md` did you last actually read before acting?\n2. Would you rather the agent forget a true thing, or remember a false one?\n3. When two remembered rules disagree, who decides — and when?\n4. What would make you delete `.memory\u002F` entirely?\n\nRationale, open decisions, field schema: [`..\u002F..\u002FDESIGN.md`](..\u002F..\u002FDESIGN.md).\n","# Agent Memory — promotion is earned, not asserted\n\n> **Portability:** stdlib only. No database, no embeddings, no network, no LLM calls.\n\n## The problem\n\nA project's `CLAUDE.md` is a memory system with one tier and no eviction: every\ndurable fact and every passing preference land in the same always-loaded file,\nuntil the important lines are diluted by the incidental ones. Facts learned\nmid-session vanish at teardown unless someone writes them down.\n\n**The fix is not more storage — it is a promotion ladder.** A claim earns its\nway toward always-loaded context by recurring; a human confirms the last step.\n\n## The four tiers\n\nTiers are distinguished by **injection policy**, not storage format.\n\n| Tier | Holds | Injected | Committed |\n|---|---|---|---|\n| **L0** | raw session transcripts | never | no (already on disk) |\n| **L1** | candidate atoms | on relevance, at prompt time | no (gitignored) |\n| **L2** | this project's context | every session start | yes, after adopt |\n| **L3** | stable cross-project persona | always | yes, after adopt |\n\n## The gates\n\nNothing moves up because it sounded important. It moves up because it recurred.\n\n- **L0 → L1** — an explicit marker fires (a directive, a correction, a stated\n  preference, a named lesson, a reproducible failure). Rule-based, high\n  precision, deliberately low recall.\n- **L1 → L2** — ≥ 3 distinct sessions spanning ≥ 2 distinct calendar days. A\n  claim stated outright needs 2 sessions; the distinct-day rule still applies. A\n  verified claim promotes on one observation and is the only day-exempt path.\n- **L2 → L3** — held in ≥ 2 distinct projects, aged ≥ 30 days, uncontested.\n\n**Two gates refuse rather than guess.** A claim whose text was altered by\nredaction never promotes on evidence alone — the flag firing is evidence the\nsource was sensitive, and a lexical filter finding one secret is not proof it\nfound all of them. A claim with an open contradiction is frozen at L1 until a\nhuman resolves it; the incumbent is never silently overwritten.\n\n## Use it\n\n```bash\n# what is remembered, and what is blocking the next promotion\npython3 scripts\u002Fmemory_inspect.py --tier L1\n\n# where did this line come from — sessions, days, transcript, quoted source\npython3 scripts\u002Fmemory_inspect.py --why \"PR base branch is dev\"\n\n# every claim with an open contradiction, both directions of the join\npython3 scripts\u002Fmemory_inspect.py --contested\n\n# dry-run the promotion pass; writes nothing\npython3 scripts\u002Fmemory_promote.py\n```\n\nThree hooks run the loop unattended: `SessionStart` injects L2 + L3,\n`UserPromptSubmit` recalls relevant L1 atoms, `SessionEnd` captures and stages.\nEach is disabled independently with `AGENT_MEMORY_SESSIONSTART=0`,\n`AGENT_MEMORY_USERPROMPTSUBMIT=0`, `AGENT_MEMORY_SESSIONEND=0`. Every hook fails\nopen: a broken memory system costs you memory, never a session.\n\n## Hard rules\n\n1. **Redact before writing.** Every atom passes the filter before it reaches\n   disk. Anything altered is quarantined from promotion.\n2. **Propose, never apply.** Promotions land in `.memory\u002Fstaged\u002F`. Only an\n   explicit `\u002Fcs:memory adopt` touches a `CLAUDE.md`, and it backs both up first.\n3. **Cite, don't invent.** Every atom carries a back-pointer to the transcript\n   line that produced it. `--why` resolving to *ambiguous* prints nothing rather\n   than guess: a wrong citation is worse than a missing one.\n4. **Never surface a contested claim as fact.** It is still injected — hiding\n   the conflict is worse — but always tagged.\n5. **The committed tiers carry no paths.** Promotion strips the back-pointer\n   prefix, which embeds an OS username.\n\n## Forcing questions\n\nWalk these one at a time before trusting the store.\n\n1. Which line in your `CLAUDE.md` did you last actually read before acting?\n2. Would you rather the agent forget a true thing, or remember a false one?\n3. When two remembered rules disagree, who decides — and when?\n4. What would make you delete `.memory\u002F` entirely?\n\nRationale, open decisions, field schema: [`..\u002F..\u002FDESIGN.md`](..\u002F..\u002FDESIGN.md).\n",{"name":5,"license":14,"metadata":47,"description":7,"argument-hint":51},{"version":48,"build_pattern":49,"distinct_from":50},"1.0.0","Tencent TencentDB-Agent-Memory's tiering concept rebuilt natively on Claude Code hooks; deterministic recurrence gates, no LLM, no database","llm-wiki (a vault you write on purpose; this writes itself from sessions); skillopt-sleep (replays tasks to improve a skill; this extracts facts to remember); memory-engineering (audits and prices any memory system; this IS one, and is a legitimate subject of that audit)","[optional: status | why \"\u003Cclaim>\" | a tier name]","main","068c2486ccc0f8d200a3db4d21ec137064eda3b4","skills\u002Fpublic-02-claude-skills-main-gemini-skills-agent-memory-skill-e67d5358","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing","https:\u002F\u002Fgithub.com\u002FJiayiV","2026-08-25 16:30:52",{"source":8,"skillName":5,"commands":59,"prompt":69,"zipAvailable":70},[60,63,66],{"tool":61,"command":62},"npx","npx skills add JiayiV\u002Fquestion_mixing --skill agent-memory",{"tool":64,"command":65},"bunx","bunx skills add JiayiV\u002Fquestion_mixing --skill agent-memory",{"tool":67,"command":68},"pnpm","pnpm dlx skills add JiayiV\u002Fquestion_mixing --skill agent-memory","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fpublic-02-claude-skills-main-gemini-skills-agent-memory-skill-e67d5358\u002FSKILL.md",true,[],[73,98,125,154,182,217],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":42},"01M6NN9PJS9NHWPDG4QP93BP8F","webmcpify","Make a web app agent-ready — propose a WebMCP tool manifest, integrate, verify in a real browser, heal; unrelated code stays untouched. Use for \"webmcpify\", \"add WebMCP\", or \"expose app actions to AI agents\".","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fwebmcpify\u002FSKILL.md",[],[75,80,81,82,83,40,84,85,86,87,88,89,90,91,92,93,94,95,96,97],"github","ai agents","web","notes","browser","https","actions","manifest","inventory","verify","tool","use","ready","unrelated","app","scope","for","webmcp",{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":102,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":42},"01M6NN6KZE5RBN6NGHN2WXV1RE","vardoger-analyze","Use when the user asks to personalize the GitHub Copilot CLI assistant, adapt Copilot to their style, use vardoger, or analyze their Copilot CLI conversation history. Reads the local session directory at `~\u002F.copilot\u002Fsession-state\u002F`, extracts recurring preferences and conventions, and writes a fenced personalization block into `~\u002F.copilot\u002Fcopilot-instructions.md`. Runs entirely on the user's machine via the local `vardoger` CLI (`pipx install vardoger`); no network calls and no uploads. Triggers: 'personalize my copilot', 'analyze my copilot history', 'tailor copilot to me', 'run vardoger', 'update my copilot instructions from history', 'make copilot learn my style'.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fvardoger-analyze\u002FSKILL.md",[],[106,107,108,109,110,111,112,113,41,114,115,116,117,118,119,120,121,122,123,124],"vardoger analyze","analyze","directory","cli","github copilot","copilot","run","update","network","my style","install","conversation","me","local","github copilot cli","pipx","my","at","copilot instructions",{"skillNo":126,"name":127,"nameZh":6,"description":128,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":129,"categories":130,"keywords":135,"publishedAt":42},"01M6NN3CZGQP3XAJ04BM1NNYGT","transloadit-media-processing","Process media files (video, audio, images, documents) using Transloadit. Use when asked to encode video to HLS\u002FMP4, generate thumbnails, resize or watermark images, extract audio, concatenate clips, add subtitles, OCR documents, or run any media processing pipeline. Covers 86+ processing robots for file transformation at scale.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Ftransloadit-media-processing\u002FSKILL.md",[131,134],{"categoryNo":132,"categoryName":133,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000016","图像与多媒体",{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[136,137,138,139,140,141,142,143,144,145,146,147,148,91,149,150,151,152,96,153],"transloadit media processing","Media","media processing","processing","MCP","mcp server","documents","subtitles","pipeline","ocr","Video","Audio","mp4","images","files","transformation","free","hls",{"skillNo":155,"name":156,"nameZh":6,"description":157,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":158,"categories":159,"keywords":163,"publishedAt":42},"01M6NN28E92F9GH4WFYQ2H1Y15","terraform-azurerm-set-diff-analyzer","Analyze Terraform plan JSON output for AzureRM Provider to distinguish between false-positive diffs (order-only changes in Set-type attributes) and actual resource changes. Use when reviewing terraform plan output for Azure resources like Application Gateway, Load Balancer, Firewall, Front Door, NSG, and other resources with Set-type attributes that cause spurious diffs due to internal ordering changes.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fterraform-azurerm-set-diff-analyzer\u002FSKILL.md",[160],{"categoryNo":161,"categoryName":162,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000011","运维与部署",[164,165,166,167,168,169,170,171,172,91,173,174,175,176,177,178,179,180,181,96],"terraform","SET","terraform azurerm set diff analyzer","diff","analyzer","diffs","azure","gateway","firewall","only","resource changes","provider","resources","load balancer","attributes","front","load","balancer",{"skillNo":183,"name":184,"nameZh":6,"description":185,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":186,"categories":187,"keywords":199,"publishedAt":42},"01M6NN1QMH2TDYWTH3XY3N5XPV","technical-job-search","Use this skill when a software engineer asks for help with job search tasks: parsing or analyzing a job description, tailoring a CV\u002Fresume, writing a cover letter, evaluating a job offer, or drafting a post-interview follow-up email. Do not activate for general career advice unrelated to an active job search action.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Ftechnical-job-search\u002FSKILL.md",[188,191,196],{"categoryNo":189,"categoryName":190,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":192,"categoryName":193,"taxonomyType":194,"taxonomyTypeDesc":195},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",{"categoryNo":197,"categoryName":198,"taxonomyType":194,"taxonomyTypeDesc":195},"01JBSEEDCAT000000000000038","人力资源",[200,201,202,203,204,205,206,207,107,208,91,209,210,211,212,213,214,215,96,216],"search","Job Search","technical job search","technical","writing","tasks","email","parsing","interview","up","engineer","general","activate","do","task","cv","letter",{"skillNo":218,"name":219,"nameZh":6,"description":220,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":221,"categories":222,"keywords":223,"publishedAt":42},"01M6NMZFA67XYS1GS09WT42235","steno-mode","Shorthand-first response compression that cuts ~40% of response tokens while preserving technical precision and exact literals. Use when the user says \"steno mode\", \"shorthand mode\", \"compressed responses\", \"token reduction\", \"brief structured output\", or invokes \u002Fsteno. Supports four compression levels: lite, brief, court, machine. Do not trigger for requests needing polished prose such as onboarding\u002Ftutorial content, stakeholder or customer-facing copy, or teaching-focused explanations.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fsteno-mode\u002FSKILL.md",[],[224,225,226,227,228,203,229,91,213,230,231,232,233,234,235,236,237,238,96,239],"steno mode","mode","customer","brief","first","tutorial","teaching","tokens","as","trigger","structured output","machine","output","court","requests","response compression"]