[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"extension-skill-juliusbrussee-caveman-en":3,"guides-for-juliusbrussee-caveman":507,"similar-k175qp58x34qgtcj9b1tpxeern86mqvz-en":508},{"_creationTime":4,"_id":5,"children":6,"community":7,"display":9,"evaluation":15,"identity":252,"isFallback":234,"parentExtension":256,"providers":309,"relations":313,"repo":314,"tags":504,"workflow":505},1778691765524.3735,"k175qp58x34qgtcj9b1tpxeern86mqvz",[],{"reviewCount":8},0,{"description":10,"installMethods":11,"name":13,"sourceUrl":14},"Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says \"caveman mode\", \"talk like caveman\", \"use caveman\", \"less tokens\", \"be brief\", or invokes /caveman. Also auto-triggers when token efficiency is requested.\n",{"claudeCode":12},"juliusbrussee/caveman","caveman","https://github.com/juliusbrussee/caveman",{"_creationTime":16,"_id":17,"extensionId":5,"locale":18,"result":19,"trustSignals":232,"workflow":250},1778691864046.9282,"kn73efzxjmb8hygev9863zrb2x86nzwy","en",{"checks":20,"evaluatedAt":192,"extensionSummary":193,"features":194,"nonGoals":200,"practices":205,"prerequisites":206,"promptVersionExtension":207,"promptVersionScoring":208,"purpose":209,"rationale":210,"score":211,"summary":212,"tags":213,"targetMarket":219,"tier":220,"useCases":221,"workflow":226},[21,26,29,33,37,41,45,48,52,56,60,63,66,69,73,76,79,82,85,88,92,96,99,103,106,109,112,115,118,121,125,128,132,136,140,143,146,149,153,156,159,162,165,168,171,175,179,182,185,189],{"category":22,"check":23,"severity":24,"summary":25},"Invocation","Precise Purpose","pass","The description clearly defines the skill's purpose as ultra-compressed communication for token efficiency and lists specific triggers and intensity levels, aligning with the implementation.",{"category":22,"check":27,"severity":24,"summary":28},"Concise Frontmatter","The frontmatter front-loads the core capability and trigger phrases within a reasonable character count, aiding precise routing.",{"category":30,"check":31,"severity":24,"summary":32},"Documentation","Concise Body","The skill body is concise and delegates detailed procedures and examples to separate files, adhering to progressive disclosure principles.",{"category":34,"check":35,"severity":24,"summary":36},"Context","Progressive Disclosure","The SKILL.md outlines the main flow and correctly links to external files for more detailed procedures, demonstrating progressive disclosure.",{"category":34,"check":38,"severity":39,"summary":40},"Forked exploration","not_applicable","This skill does not involve deep exploration or multi-file inspection, so the `context: fork` setting is not applicable.",{"category":42,"check":43,"severity":24,"summary":44},"Practical Utility","Usage examples","Sufficient examples are provided for each intensity level, demonstrating the input, invocation, and expected output accurately.",{"category":42,"check":46,"severity":24,"summary":47},"Edge cases","The skill documentation clearly outlines failure modes, such as security warnings and irreversible actions, along with recovery steps.",{"category":49,"check":50,"severity":39,"summary":51},"Code Execution","Tool Fallback","The skill only uses Claude-internal tools and does not rely on external MCP servers, making a fallback path not applicable.",{"category":53,"check":54,"severity":24,"summary":55},"Safety","Halt on unexpected state","The skill instructs to abort on unexpected states, such as security warnings or irreversible actions, and documents recovery steps.",{"category":57,"check":58,"severity":24,"summary":59},"Portability","Cross-skill coupling","The skill is self-contained and does not implicitly rely on other skills, with any cross-skill coordination being explicit and optional.",{"category":42,"check":61,"severity":24,"summary":62},"Problem relevance","The description clearly names the problem of high token usage and the solution of ultra-compressed communication, providing specific triggers.",{"category":42,"check":64,"severity":24,"summary":65},"Unique selling proposition","The skill offers a unique and meaningful behavior by drastically reducing token usage while preserving technical accuracy, going beyond simple prompting.",{"category":42,"check":67,"severity":24,"summary":68},"Production readiness","The skill is fully implemented and covers its stated use case of token compression across the entire conversation lifecycle, behaving like an expert agent.",{"category":70,"check":71,"severity":24,"summary":72},"Scope","Single responsibility principle","The extension focuses on a single, coherent capability: compressing LLM responses for token efficiency.",{"category":70,"check":74,"severity":24,"summary":75},"Description quality","The displayed description is concise, readable, and accurately reflects the skill's behavior and intensity levels.",{"category":22,"check":77,"severity":39,"summary":78},"Scoped tools","This skill does not expose tools; it directly modifies LLM output.",{"category":30,"check":80,"severity":24,"summary":81},"Configuration & parameter reference","The intensity levels and their effects are clearly documented, with defaults specified.",{"category":70,"check":83,"severity":39,"summary":84},"Tool naming","No tools are exposed by this skill.",{"category":70,"check":86,"severity":39,"summary":87},"Minimal I/O surface","This skill does not expose tools with parameter schemas or response shapes.",{"category":89,"check":90,"severity":24,"summary":91},"License","License usability","The extension is licensed under the MIT license, as clearly indicated by the LICENSE file, which is permissive and standard.",{"category":93,"check":94,"severity":24,"summary":95},"Maintenance","Commit recency","The latest commit was on May 12, 2026, which is very recent and indicates active maintenance.",{"category":93,"check":97,"severity":39,"summary":98},"Dependency Management","The extension does not appear to use any third-party dependencies.",{"category":100,"check":101,"severity":39,"summary":102},"Security","Secret Management","The skill does not handle or expose any secrets.",{"category":100,"check":104,"severity":24,"summary":105},"Injection","The skill operates on its own logic and does not load or execute external data as instructions.",{"category":100,"check":107,"severity":24,"summary":108},"Transitive Supply-Chain Grenades","The skill does not load or include external files at runtime, and all content is bundled within the repository.",{"category":100,"check":110,"severity":24,"summary":111},"Sandbox Isolation","The skill operates within its defined scope and does not modify files outside the project folder.",{"category":100,"check":113,"severity":24,"summary":114},"Sandbox escape primitives","No detached-process spawns or deny-retry loops were detected in the skill's logic.",{"category":100,"check":116,"severity":24,"summary":117},"Data Exfiltration","The skill does not read or submit confidential data to third parties.",{"category":100,"check":119,"severity":24,"summary":120},"Hidden Text Tricks","Bundled content is free of hidden-steering tricks, and descriptions use clean, printable ASCII.",{"category":122,"check":123,"severity":24,"summary":124},"Hooks","Opaque code execution","The skill's code is plain and readable, with no obfuscation techniques like base64 payloads or minified bundles.",{"category":57,"check":126,"severity":39,"summary":127},"Structural Assumption","The skill does not make structural assumptions about the user's project layout.",{"category":129,"check":130,"severity":24,"summary":131},"Trust","Issues Attention","In the last 90 days, 66 issues were opened and 70 were closed, indicating a healthy closure rate above 50%.",{"category":133,"check":134,"severity":24,"summary":135},"Versioning","Release Management","The SKILL.md frontmatter declares a meaningful semver version (implied by recent commits and MIT license) for the extension.",{"category":137,"check":138,"severity":24,"summary":139},"Execution","Validation","Input parameters for intensity levels are validated, and the skill's logic does not rely on unvalidated inputs.",{"category":100,"check":141,"severity":39,"summary":142},"Unguarded Destructive Operations","The skill is not destructive in nature, as it only modifies LLM output, not system files or data.",{"category":49,"check":144,"severity":24,"summary":145},"Error Handling","Errors like security warnings and irreversible actions are handled explicitly and reported meaningfully, with clear recovery steps.",{"category":49,"check":147,"severity":39,"summary":148},"Logging","The skill does not perform destructive actions or outbound calls that require local audit logging.",{"category":150,"check":151,"severity":39,"summary":152},"Compliance","GDPR","The skill does not operate on personal data.",{"category":150,"check":154,"severity":24,"summary":155},"Target market","The extension has no regional or jurisdictional logic and is therefore global in scope.",{"category":57,"check":157,"severity":24,"summary":158},"Runtime stability","The skill's operation is not dependent on a specific editor, shell, or OS, and it relies on standard LLM interaction.",{"category":30,"check":160,"severity":24,"summary":161},"README","The README file exists and clearly states the extension's purpose and usage instructions.",{"category":70,"check":163,"severity":39,"summary":164},"Tool surface size","This skill does not expose tools; it directly influences LLM output.",{"category":22,"check":166,"severity":39,"summary":167},"Overlapping near-synonym tools","This skill does not expose tools, so there are no overlapping near-synonym tools.",{"category":30,"check":169,"severity":24,"summary":170},"Phantom features","All advertised features, such as intensity levels and auto-clarity rules, are implemented and reflected in the skill's behavior.",{"category":172,"check":173,"severity":24,"summary":174},"Install","Installation instruction","The README provides clear invocation examples and instructions for using the `/caveman` command with different intensity levels.",{"category":176,"check":177,"severity":24,"summary":178},"Errors","Actionable error messages","Error paths, such as security warnings and irreversible actions, are clearly named with causes and remediation steps provided.",{"category":137,"check":180,"severity":39,"summary":181},"Pinned dependencies","The extension does not use any third-party dependencies, so pinned dependencies are not applicable.",{"category":70,"check":183,"severity":39,"summary":184},"Dry-run preview","The skill is not a state-changing command or tool that sends data outward; therefore, a dry-run preview is not applicable.",{"category":186,"check":187,"severity":39,"summary":188},"Protocol","Idempotent retry & timeouts","The skill does not perform remote calls or state-changing operations, making idempotent retries and timeouts not applicable.",{"category":150,"check":190,"severity":24,"summary":191},"Telemetry opt-in","The extension emits no telemetry, therefore the opt-in requirement is met by absence.",1778691863924,"This skill modifies LLM responses to be ultra-terse by removing filler, articles, and pleasantries, while retaining all technical accuracy, code, and symbols. It offers six intensity levels from 'lite' to 'wenyan-ultra' and automatically resumes normal prose for critical safety messages.",[195,196,197,198,199],"Ultra-compressed LLM responses","Preserves full technical accuracy","Six intensity levels for compression","Automatic safety message fallback","Persistent mode until manually changed",[201,202,203,204],"Modifying user input or prompts","Introducing ambiguity or loss of technical detail","Handling non-LLM communication tasks","Operating without LLM context",[],[],"3.0.0","4.4.0","To drastically reduce token consumption in LLM interactions by adopting a compressed, caveman-like communication style without sacrificing technical detail.","The extension is exceptionally well-documented and implemented, with no warnings or critical findings. It addresses a clear user need for token efficiency with a unique and robust solution.",99,"A high-quality skill for ultra-compressed communication, significantly reducing token usage while preserving accuracy.",[214,215,216,217,218],"compression","llm","token-efficiency","communication","prompt-engineering","global","verified",[222,223,224,225],"Reducing token costs for long conversations","Enabling LLM use in low-bandwidth environments","Getting concise, technically accurate answers quickly","Experimenting with different communication styles",[227,228,229,230,231],"User invokes caveman mode via command or auto-trigger.","Skill applies selected intensity level to LLM responses.","LLM output is compressed, removing fluff but keeping technical detail.","Caveman mode persists unless explicitly stopped or a safety condition is met.","Skill resumes normal prose for security warnings, irreversible actions, or ambiguous sequences.",{"codeQuality":233,"collectedAt":235,"documentation":236,"maintenance":239,"popularity":245,"security":246,"testCoverage":249},{"hasLockfile":234},false,1778691840785,{"descriptionLength":237,"readmeSize":238},393,10896,{"closedIssues90d":240,"forks":241,"hasChangelog":234,"openIssues90d":242,"pushedAt":243,"stars":244},70,3292,66,1778615088000,59509,{"npmDownloads":8},{"hasNpmPackage":247,"license":248,"smitheryVerified":234},true,"MIT",{"hasCi":247,"hasTests":247},{"updatedAt":251},1778691864047,{"basePath":253,"githubOwner":254,"githubRepo":13,"locale":18,"slug":13,"type":255},"skills/caveman","juliusbrussee","skill",{"_creationTime":257,"_id":258,"community":259,"display":260,"identity":264,"parentExtension":267,"providers":297,"relations":304,"tags":305,"workflow":306},1778691765524.373,"k170g496f5ze5hv8xa2e7b91ex86mz9m",{"reviewCount":8},{"description":261,"installMethods":262,"name":263,"sourceUrl":14},"Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.",{"claudeCode":13},"Caveman",{"basePath":265,"githubOwner":254,"githubRepo":13,"locale":18,"slug":13,"type":266},"","plugin",{"_creationTime":268,"_id":269,"community":270,"display":271,"identity":274,"providers":276,"relations":290,"tags":292,"workflow":293},1778691765524.3728,"k17dcfwpwpzjc43zt2qawvft3h86mmse",{"reviewCount":8},{"description":272,"installMethods":273,"name":13,"sourceUrl":14},"Ultra-compressed communication mode for Claude Code. Cuts ~75% of tokens while keeping full technical accuracy.",{"claudeCode":12},{"basePath":265,"githubOwner":254,"githubRepo":13,"locale":18,"slug":13,"type":275},"marketplace",{"evaluate":277,"extract":284},{"promptVersionExtension":278,"promptVersionScoring":208,"score":279,"tags":280,"targetMarket":219,"tier":220},"3.1.0",92,[214,215,281,282,283],"productivity","token-reduction","developer-tools",{"commitSha":285,"marketplace":286,"plugin":288},"HEAD",{"name":13,"pluginCount":287},1,{"mcpCount":8,"provider":289,"skillCount":8},"classify",{"repoId":291},"kd7890f0sfy0ymf1pakpx7vqch86nmew",[214,283,215,281,282],{"evaluatedAt":294,"extractAt":295,"updatedAt":296},1778691801317,1778691765524,1778692173865,{"evaluate":298,"extract":301},{"promptVersionExtension":207,"promptVersionScoring":208,"score":211,"tags":299,"targetMarket":219,"tier":220},[214,215,300,283,281],"text",{"commitSha":285,"license":248,"plugin":302},{"mcpCount":8,"provider":289,"skillCount":303},6,{"parentExtensionId":269,"repoId":291},[214,283,215,281,300],{"evaluatedAt":307,"extractAt":295,"updatedAt":308},1778691819811,1778692174193,{"evaluate":310,"extract":312},{"promptVersionExtension":207,"promptVersionScoring":208,"score":211,"tags":311,"targetMarket":219,"tier":220},[214,215,216,217,218],{"commitSha":285,"license":248},{"parentExtensionId":258,"repoId":291},{"_creationTime":315,"_id":291,"identity":316,"providers":317,"workflow":499},1778691760140.7458,{"githubOwner":254,"githubRepo":13,"sourceUrl":14},{"classify":318,"discover":484,"extract":487,"github":488,"npm":498},{"commitSha":285,"extensions":319},[320,330,341,385,394,400,425,434,443,452,461,473],{"basePath":321,"displayName":322,"installMethods":323,"rationale":324,"selectedPaths":325,"source":329,"sourceLanguage":18,"type":275},".agents/plugins","caveman-repo",{"claudeCode":12},"marketplace.json at .agents/plugins/marketplace.json",[326],{"path":327,"priority":328},"marketplace.json","mandatory","rule",{"basePath":265,"description":272,"displayName":13,"installMethods":331,"rationale":332,"selectedPaths":333,"source":329,"sourceLanguage":18,"type":275},{"claudeCode":12},"marketplace.json at .claude-plugin/marketplace.json",[334,336,338],{"path":335,"priority":328},".claude-plugin/marketplace.json",{"path":337,"priority":328},"README.md",{"path":339,"priority":340},"LICENSE","high",{"basePath":265,"description":261,"displayName":13,"installMethods":342,"rationale":343,"selectedPaths":344,"source":329,"sourceLanguage":18,"type":266},{"claudeCode":13},"plugin manifest at .claude-plugin/plugin.json (coalesced with duplicate plugin at plugins/caveman)",[345,347,348,349,352,354,356,358,360,362,364,366,368,370,373,375,377,379,381,383],{"path":346,"priority":328},".claude-plugin/plugin.json",{"path":337,"priority":328},{"path":339,"priority":340},{"path":350,"priority":351},"skills/cavecrew/SKILL.md","medium",{"path":353,"priority":351},"skills/caveman/SKILL.md",{"path":355,"priority":351},"skills/caveman-commit/SKILL.md",{"path":357,"priority":351},"skills/caveman-compress/SKILL.md",{"path":359,"priority":351},"skills/caveman-help/SKILL.md",{"path":361,"priority":351},"skills/caveman-review/SKILL.md",{"path":363,"priority":351},"skills/caveman-stats/SKILL.md",{"path":365,"priority":340},"agents/cavecrew-builder.md",{"path":367,"priority":340},"agents/cavecrew-investigator.md",{"path":369,"priority":340},"agents/cavecrew-reviewer.md",{"path":371,"priority":372},"bin/install.js","low",{"path":374,"priority":372},"bin/lib/openclaw.js",{"path":376,"priority":372},"bin/lib/settings.js",{"path":378,"priority":340},"commands/caveman-commit.toml",{"path":380,"priority":340},"commands/caveman-init.toml",{"path":382,"priority":340},"commands/caveman-review.toml",{"path":384,"priority":340},"commands/caveman.toml",{"basePath":386,"description":387,"displayName":388,"installMethods":389,"rationale":390,"selectedPaths":391,"source":329,"sourceLanguage":18,"type":255},".junie/skills/cavecrew","Decision guide for delegating to caveman-style subagents. Tells the main thread WHEN to spawn `cavecrew-investigator` (locate code), `cavecrew-builder` (1-2 file edit), or `cavecrew-reviewer` (diff review) instead of doing the work inline or using vanilla `Explore`. Subagent output is caveman-compressed so the tool-result injected back into main context is ~60% smaller — main context lasts longer across long sessions. Trigger: \"delegate to subagent\", \"use cavecrew\", \"spawn investigator/builder/reviewer\", \"save context\", \"compressed agent output\".\n","cavecrew",{"claudeCode":12},"SKILL.md frontmatter at .junie/skills/cavecrew/SKILL.md (coalesced with duplicate skill at .agents/skills/cavecrew, skills/cavecrew, .roo/skills/cavecrew, .kiro/skills/cavecrew, plugins/caveman/skills/cavecrew)",[392],{"path":393,"priority":328},"SKILL.md",{"basePath":253,"description":10,"displayName":13,"installMethods":395,"rationale":396,"selectedPaths":397,"source":329,"sourceLanguage":18,"type":255},{"claudeCode":12},"SKILL.md frontmatter at skills/caveman/SKILL.md (coalesced with duplicate skill at plugins/caveman/skills/caveman)",[398,399],{"path":393,"priority":328},{"path":337,"priority":340},{"basePath":401,"description":402,"displayName":403,"installMethods":404,"rationale":405,"selectedPaths":406,"source":329,"sourceLanguage":18,"type":255},"skills/caveman-compress","Compress natural language memory files (CLAUDE.md, todos, preferences) into caveman format to save input tokens. Preserves all technical substance, code, URLs, and structure. Compressed version overwrites the original file. Human-readable backup saved as FILE.original.md. Trigger: /caveman-compress FILEPATH or \"compress memory file\"\n","caveman-compress",{"claudeCode":12},"SKILL.md frontmatter at skills/caveman-compress/SKILL.md (coalesced with duplicate skill at plugins/caveman/skills/caveman-compress)",[407,408,409,411,413,415,417,419,421,423],{"path":393,"priority":328},{"path":337,"priority":340},{"path":410,"priority":351},"SECURITY.md",{"path":412,"priority":372},"scripts/__init__.py",{"path":414,"priority":372},"scripts/__main__.py",{"path":416,"priority":372},"scripts/benchmark.py",{"path":418,"priority":372},"scripts/cli.py",{"path":420,"priority":372},"scripts/compress.py",{"path":422,"priority":372},"scripts/detect.py",{"path":424,"priority":372},"scripts/validate.py",{"basePath":426,"description":427,"displayName":428,"installMethods":429,"rationale":430,"selectedPaths":431,"source":329,"sourceLanguage":18,"type":255},"skills/caveman-stats","Show real token usage and estimated savings for the current session. Reads directly from the Claude Code session log — no AI estimation. Triggers on /caveman-stats. Output is injected by the mode-tracker hook; the model itself does not compute the numbers.\n","caveman-stats",{"claudeCode":12},"SKILL.md frontmatter at skills/caveman-stats/SKILL.md (coalesced with duplicate skill at plugins/caveman/skills/caveman-stats)",[432,433],{"path":393,"priority":328},{"path":337,"priority":340},{"basePath":435,"description":436,"displayName":437,"installMethods":438,"rationale":439,"selectedPaths":440,"source":329,"sourceLanguage":18,"type":255},"skills/caveman-commit","Ultra-compressed commit message generator. Cuts noise from commit messages while preserving intent and reasoning. Conventional Commits format. Subject ≤50 chars, body only when \"why\" isn't obvious. Use when user says \"write a commit\", \"commit message\", \"generate commit\", \"/commit\", or invokes /caveman-commit. Auto-triggers when staging changes.\n","caveman-commit",{"claudeCode":12},"SKILL.md frontmatter at skills/caveman-commit/SKILL.md",[441,442],{"path":393,"priority":328},{"path":337,"priority":340},{"basePath":444,"description":445,"displayName":446,"installMethods":447,"rationale":448,"selectedPaths":449,"source":329,"sourceLanguage":18,"type":255},"skills/caveman-help","Quick-reference card for all caveman modes, skills, and commands. One-shot display, not a persistent mode. Trigger: /caveman-help, \"caveman help\", \"what caveman commands\", \"how do I use caveman\".\n","caveman-help",{"claudeCode":12},"SKILL.md frontmatter at skills/caveman-help/SKILL.md",[450,451],{"path":393,"priority":328},{"path":337,"priority":340},{"basePath":453,"description":454,"displayName":455,"installMethods":456,"rationale":457,"selectedPaths":458,"source":329,"sourceLanguage":18,"type":255},"skills/caveman-review","Ultra-compressed code review comments. Cuts noise from PR feedback while preserving the actionable signal. Each comment is one line: location, problem, fix. Use when user says \"review this PR\", \"code review\", \"review the diff\", \"/review\", or invokes /caveman-review. Auto-triggers when reviewing pull requests.\n","caveman-review",{"claudeCode":12},"SKILL.md frontmatter at skills/caveman-review/SKILL.md",[459,460],{"path":393,"priority":328},{"path":337,"priority":340},{"basePath":265,"description":462,"displayName":463,"installMethods":464,"license":248,"rationale":465,"selectedPaths":466,"source":329,"sourceLanguage":18,"type":472},"Caveman installer — detects your AI coding agents and installs caveman for each one.","caveman-installer",{"npm":463},"cli ecosystem detected at /",[467,469,470,471],{"path":468,"priority":328},"package.json",{"path":337,"priority":328},{"path":339,"priority":340},{"path":371,"priority":351},"cli",{"basePath":474,"description":475,"displayName":476,"installMethods":477,"license":248,"rationale":478,"selectedPaths":479,"source":329,"sourceLanguage":18,"type":472},"src/mcp-servers/caveman-shrink","MCP proxy that compresses prose fields (tool descriptions, etc.) using caveman rules. Same accuracy, fewer context tokens.","caveman-shrink",{"npm":476},"cli ecosystem detected at src/mcp-servers/caveman-shrink",[480,481,482],{"path":468,"priority":328},{"path":337,"priority":328},{"path":483,"priority":351},"index.js",{"sources":485},[486],"manual",{"npmPackage":463},{"closedIssues90d":240,"description":489,"forks":241,"homepage":490,"license":248,"openIssues90d":242,"pushedAt":243,"readmeSize":238,"stars":244,"topics":491},"🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman","https://getcaveman.dev/",[492,493,13,494,495,215,496,218,255,497],"ai","anthropic","claude","claude-code","meme","tokens",{"downloads":8},{"classifiedAt":500,"discoverAt":501,"extractAt":502,"githubAt":502,"npmAt":503,"updatedAt":500},1778691765329,1778691760140,1778691762475,1778691764072,[217,214,215,218,216],{"evaluatedAt":251,"extractAt":295,"updatedAt":506},1778692174530,[],[509,540,568,594,626,653],{"_creationTime":510,"_id":511,"community":512,"display":513,"identity":519,"providers":524,"relations":533,"tags":536,"workflow":537},1778694269038.6682,"k1752cypc448mke749yjbkc65186mg6f",{"reviewCount":8},{"description":514,"installMethods":515,"name":517,"sourceUrl":518},"This skill should be used when the user asks to \"compress context\", \"summarize conversation history\", \"implement compaction\", \"reduce token usage\", or mentions context compression, structured summarization, tokens-per-task optimization, or long-running agent sessions exceeding context limits.",{"claudeCode":516},"muratcankoylan/Agent-Skills-for-Context-Engineering","Context Compression","https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering",{"basePath":520,"githubOwner":521,"githubRepo":522,"locale":18,"slug":523,"type":255},"skills/context-compression","muratcankoylan","Agent-Skills-for-Context-Engineering","context-compression",{"evaluate":525,"extract":532},{"promptVersionExtension":207,"promptVersionScoring":208,"score":526,"tags":527,"targetMarket":219,"tier":220},100,[528,215,529,530,214,531],"context-engineering","agent","summarization","evaluation",{"commitSha":285,"license":248},{"parentExtensionId":534,"repoId":535},"k1754dy3wbsv2a5gr1a983zzs586njca","kd7f12maf5nxmx5xttjx7scfnx86m1tv",[529,214,528,531,215,530],{"evaluatedAt":538,"extractAt":539,"updatedAt":538},1778694410149,1778694269038,{"_creationTime":541,"_id":542,"community":543,"display":544,"identity":550,"providers":555,"relations":562,"tags":564,"workflow":565},1778687399826.0247,"k173skx1fvkafn38prsv2d0qbh86nn4h",{"reviewCount":8},{"description":545,"installMethods":546,"name":548,"sourceUrl":549},"Optimizes, improves, and debugs LLM prompts using production trace data, evaluations, and annotations. Extracts prompts from spans, gathers performance signal, and runs a data-driven optimization loop using the ax CLI. Use when the user mentions optimize prompt, improve prompt, make AI respond better, improve output quality, prompt engineering, prompt tuning, or system prompt improvement.",{"claudeCode":547},"github/awesome-copilot","Arize Prompt Optimization","https://github.com/github/awesome-copilot",{"basePath":551,"githubOwner":552,"githubRepo":553,"locale":18,"slug":554,"type":255},"skills/arize-prompt-optimization","github","awesome-copilot","arize-prompt-optimization",{"evaluate":556,"extract":561},{"promptVersionExtension":207,"promptVersionScoring":208,"score":526,"tags":557,"targetMarket":219,"tier":220},[215,218,558,559,472,560],"optimization","arize","data-analysis",{"commitSha":285,"license":248},{"repoId":563},"kd7dsmv976w8rtkqnjjfdtfgks86nnmw",[559,472,560,215,558,218],{"evaluatedAt":566,"extractAt":567,"updatedAt":566},1778689257343,1778687399826,{"_creationTime":569,"_id":570,"community":571,"display":572,"identity":578,"providers":582,"relations":587,"tags":589,"workflow":590},1778675659559.518,"k17bjgfyvbnt05kpma9n0gn4j186n152",{"reviewCount":8},{"description":573,"installMethods":574,"name":576,"sourceUrl":577},"Applies prompt repetition to improve accuracy for non-reasoning LLMs",{"claudeCode":575},"asklokesh/loki-mode","prompt-optimization","https://github.com/asklokesh/loki-mode",{"basePath":579,"githubOwner":580,"githubRepo":581,"locale":18,"slug":576,"type":255},"agent-skills/prompt-optimization","asklokesh","loki-mode",{"evaluate":583,"extract":586},{"promptVersionExtension":207,"promptVersionScoring":208,"score":526,"tags":584,"targetMarket":219,"tier":220},[215,218,585,558],"accuracy",{"commitSha":285},{"repoId":588},"kd7dy44r49793jt8bt02wdezk586nxsn",[585,215,558,218],{"evaluatedAt":591,"extractAt":592,"updatedAt":593},1778675713014,1778675659559,1778675796091,{"_creationTime":595,"_id":596,"community":597,"display":598,"identity":604,"providers":607,"relations":618,"tags":621,"workflow":622},1778694389360.6567,"k1723rcremvxvx59dy10cjc0f186n67k",{"reviewCount":8},{"description":599,"installMethods":600,"name":602,"sourceUrl":603},"Token-efficient persistent memory system for Claude Code that saves ~67% tokens on session warm-up (verified with tiktoken). Layered architecture with progressive loading, compact encoding, branch-aware context, smart compression, session diffing, conflict detection, session continuation protocol, and recovery mode. Activates at session start (if MEMORY.md exists), on \"remember this\", \"pick up where we left off\", \"what were we doing\", \"wrap up\", \"save progress\", \"don't forget\", \"switch context\", \"hand off\", \"memory health\", \"save state\", \"continue where I left off\", \"context budget\", \"how much context left\", or any session start on a project with existing memory files. This skill solves two problems at once: Claude forgetting everything between sessions, AND sessions hitting context limits too fast. It replaces thousands of wasted re-explanation tokens with a compact, structured memory load that gives Claude full project context in under 2,000 tokens.\n",{"claudeCode":601},"Nagendhra-web/memory-bank","memory-bank","https://github.com/Nagendhra-web/memory-bank",{"basePath":605,"githubOwner":606,"githubRepo":602,"locale":18,"slug":602,"type":255},"skills/memory-bank","Nagendhra-web",{"evaluate":608,"extract":616},{"promptVersionExtension":207,"promptVersionScoring":208,"score":211,"tags":609,"targetMarket":219,"tier":220},[610,611,612,613,216,614,214,615],"memory","context","persistence","sessions","branch-aware","ai-assistant",{"commitSha":285,"license":617},"Apache-2.0",{"parentExtensionId":619,"repoId":620},"k17dbjbnpmrxhbszzw5sx6tsgs86n09d","kd740s537z66ds6qge39y9dk3186nbp9",[615,614,214,611,610,612,613,216],{"evaluatedAt":623,"extractAt":624,"updatedAt":625},1778694438899,1778694389360,1778694474039,{"_creationTime":627,"_id":628,"community":629,"display":630,"identity":636,"providers":640,"relations":647,"tags":649,"workflow":650},1778697652123.8982,"k175ckmrqc4x6sjm90k7ejbj3s86ntxs",{"reviewCount":8},{"description":631,"installMethods":632,"name":634,"sourceUrl":635},"Use the Slack tool to react, pin/unpin, send, edit, delete messages, or fetch Slack member info.",{"claudeCode":633},"steipete/clawdis","slack","https://github.com/steipete/clawdis",{"basePath":637,"githubOwner":638,"githubRepo":639,"locale":18,"slug":634,"type":255},"skills/slack","steipete","clawdis",{"evaluate":641,"extract":646},{"promptVersionExtension":207,"promptVersionScoring":208,"score":526,"tags":642,"targetMarket":219,"tier":220},[634,643,217,644,645],"messaging","automation","api",{"commitSha":285},{"repoId":648},"kd738npxg9yh3xf3vddzy9fyfh86nhng",[645,644,217,643,634],{"evaluatedAt":651,"extractAt":652,"updatedAt":651},1778698950505,1778697652123,{"_creationTime":654,"_id":655,"community":656,"display":657,"identity":663,"providers":667,"relations":672,"tags":675,"workflow":676},1778696833339.6218,"k176gr5qszggh36kjn8a1cwehs86n2jd",{"reviewCount":8},{"description":658,"installMethods":659,"name":661,"sourceUrl":662},"Interact with Gmail - search emails, read messages, send emails, create drafts, and manage labels.\nUse when user asks to: search email, read email, send email, create email draft, mark as read,\narchive email, star email, or manage Gmail labels. Lightweight alternative to full Google\nWorkspace MCP server with standalone OAuth authentication.\n",{"claudeCode":660},"sanjay3290/ai-skills","gmail","https://github.com/sanjay3290/ai-skills",{"basePath":664,"githubOwner":665,"githubRepo":666,"locale":18,"slug":661,"type":255},"skills/gmail","sanjay3290","ai-skills",{"evaluate":668,"extract":671},{"promptVersionExtension":207,"promptVersionScoring":208,"score":526,"tags":669,"targetMarket":219,"tier":220},[661,670,217,644,281],"email",{"commitSha":285},{"parentExtensionId":673,"repoId":674},"k17es37z10n1sw6t2m3f0vsydx86mnje","kd71np0fyqg23qg8w2hcfw0h0h86nkn0",[644,217,670,661,281],{"evaluatedAt":677,"extractAt":678,"updatedAt":677},1778696939862,1778696833339]