[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"extension-marketplace-hilyfux-knowledge-graph-en":3,"guides-for-hilyfux-knowledge-graph":345,"similar-k17cbhxz59f98xy0xnhtgaq59186n3fa-en":346},{"_creationTime":4,"_id":5,"children":6,"community":49,"display":50,"evaluation":55,"identity":265,"isFallback":252,"parentExtension":267,"providers":268,"relations":275,"repo":276,"tags":342,"workflow":343},1778688381229.894,"k17cbhxz59f98xy0xnhtgaq59186n3fa",[7],{"_creationTime":8,"_id":9,"community":10,"display":12,"identity":17,"providers":22,"relations":42,"tags":44,"workflow":45},1778688381229.8943,"k1716hqxxq72s4623edvj0sc2s86nrvv",{"reviewCount":11},0,{"description":13,"installMethods":14,"name":15,"sourceUrl":16},"Persistent memory layer for Claude Code — tracks file operations, mines co-change patterns, generates distributed CLAUDE.md knowledge nodes. Zero dependencies beyond jq.",{"claudeCode":15},"knowledge-graph","https://github.com/hilyfux/knowledge-graph",{"basePath":18,"githubOwner":19,"githubRepo":15,"locale":20,"slug":15,"type":21},"","hilyfux","en","plugin",{"evaluate":23,"extract":36},{"promptVersionExtension":24,"promptVersionScoring":25,"score":26,"tags":27,"targetMarket":34,"tier":35},"3.0.0","4.4.0",94,[15,28,29,30,31,32,33],"memory","git-native","claude-code","mcp","bash","jq","global","community",{"commitSha":37,"license":38,"plugin":39},"HEAD","MIT",{"mcpCount":11,"provider":40,"skillCount":41},"classify",1,{"parentExtensionId":5,"repoId":43},"kd726s6dkwq1hy5se33xvs6mqx86m67m",[32,30,29,33,15,31,28],{"evaluatedAt":46,"extractAt":47,"updatedAt":48},1778688431684,1778688381229,1778688497232,{"reviewCount":11},{"description":51,"installMethods":52,"name":54,"sourceUrl":16},"Persistent memory layer for Claude Code with zero-interrupt architecture",{"claudeCode":53},"hilyfux/knowledge-graph","Knowledge Graph",{"_creationTime":56,"_id":57,"extensionId":5,"locale":20,"result":58,"trustSignals":250,"workflow":263},1778688415198.2397,"kn7e4pebad85kgtktqk0m1cg7d86mmhy",{"checks":59,"evaluatedAt":206,"extensionSummary":207,"features":208,"nonGoals":216,"practices":221,"prerequisites":225,"promptVersionExtension":227,"promptVersionScoring":25,"purpose":228,"rationale":229,"score":230,"summary":231,"tags":232,"targetMarket":34,"tier":35,"useCases":237,"workflow":242},[60,65,68,71,75,78,82,87,90,93,97,101,104,108,111,114,117,120,123,126,130,134,139,143,147,150,153,156,160,163,166,169,172,175,178,182,186,190,193,197,200,203],{"category":61,"check":62,"severity":63,"summary":64},"Practical Utility","Problem relevance","pass","The description clearly states the problem of AI agents forgetting context between sessions and offers a solution with a persistent memory layer.",{"category":61,"check":66,"severity":63,"summary":67},"Unique selling proposition","The extension offers a unique approach to AI memory using git-native storage and bash/jq, avoiding common dependencies like vector stores or external services.",{"category":61,"check":69,"severity":63,"summary":70},"Production readiness","The extension provides a complete lifecycle for memory management, including hooks for Claude Code and MCP tools for other agents, with installation instructions and cross-platform support.",{"category":72,"check":73,"severity":63,"summary":74},"Scope","Single responsibility principle","The extension focuses on providing a memory layer for AI coding agents, with clear scope around file operations, git history, and context management.",{"category":72,"check":76,"severity":63,"summary":77},"Description quality","The displayed description accurately reflects the core functionality of providing persistent memory for AI coding agents.",{"category":79,"check":80,"severity":63,"summary":81},"Invocation","Scoped tools","The MCP server exposes specific, verb-noun tools like `kg_query` and `kg_read_node`, avoiding a single generalist execution tool.",{"category":83,"check":84,"severity":85,"summary":86},"Documentation","Configuration & parameter reference","info","While the README mentions configuration via env vars and tuning, specific details on all options, defaults, and precedence are not explicitly documented beyond general pointers.",{"category":72,"check":88,"severity":63,"summary":89},"Tool naming","MCP tools like `kg_status`, `kg_query`, `kg_read_node` are descriptive and follow kebab-case naming conventions.",{"category":72,"check":91,"severity":63,"summary":92},"Minimal I/O surface","The tool descriptions and resource access patterns suggest focused inputs and outputs relevant to the specific tasks of memory management and querying.",{"category":94,"check":95,"severity":63,"summary":96},"License","License usability","The project is licensed under MIT and includes a LICENSE file, which is a standard and permissive open-source license.",{"category":98,"check":99,"severity":63,"summary":100},"Maintenance","Commit recency","The last commit was on May 6, 2026, which is within the last 3 months.",{"category":98,"check":102,"severity":63,"summary":103},"Dependency Management","The extension has minimal external dependencies, primarily relying on `bash` and `jq`, which are standard system tools, and `git` which is optional.",{"category":105,"check":106,"severity":63,"summary":107},"Security","Secret Management","The extension's purpose is to manage project context and code memory, not sensitive secrets, and it operates locally within the user's repository.",{"category":105,"check":109,"severity":63,"summary":110},"Injection","The extension operates on local files and git history, and its core logic is in bash/jq scripts, with no indication of loading or executing untrusted external content.",{"category":105,"check":112,"severity":63,"summary":113},"Transitive Supply-Chain Grenades","The extension's installation script is fetched via curl, but it's a bash script for local setup and does not execute arbitrary remote code after installation. All core logic is bundled.",{"category":105,"check":115,"severity":63,"summary":116},"Sandbox Isolation","The extension operates within the user's project directory and leverages standard system tools, with no indications of attempting to modify files outside its scope or the user's project.",{"category":105,"check":118,"severity":63,"summary":119},"Sandbox escape primitives","The bash scripts for hooks and tools do not appear to contain primitives like `nohup` or detached process spawns for escaping sandbox environments.",{"category":105,"check":121,"severity":63,"summary":122},"Data Exfiltration","The extension's function is to manage local project knowledge and does not involve sending confidential data to third parties.",{"category":105,"check":124,"severity":63,"summary":125},"Hidden Text Tricks","The bundled markdown files and scripts appear to be free of hidden steering tricks, control characters, or invisible unicode sequences.",{"category":127,"check":128,"severity":63,"summary":129},"Hooks","Opaque code execution","The core logic is implemented in readable bash and jq scripts, with no evidence of obfuscation, base64 payloads, or runtime code fetching.",{"category":131,"check":132,"severity":63,"summary":133},"Portability","Structural Assumption","The installation script targets a user-specified path, and the extension generates files within the project, making reasonable assumptions about project structure.",{"category":135,"check":136,"severity":137,"summary":138},"Trust","Issues Attention","warning","There is 1 open issue and 0 closed issues in the last 90 days, indicating a low rate of issue closure and potentially slow maintainer response.",{"category":140,"check":141,"severity":63,"summary":142},"Versioning","Release Management","The plugin has a `version` field in `marketplace.json` set to '1.2.0' and a CHANGELOG.md file is present.",{"category":144,"check":145,"severity":85,"summary":146},"Code Execution","Validation","While the bash scripts likely handle arguments, there's no explicit mention or evidence of a schema validation library like Zod or pydantic being used for input sanitization.",{"category":105,"check":148,"severity":63,"summary":149},"Unguarded Destructive Operations","The extension primarily manages knowledge and context, with no explicit destructive operations like file deletion or forceful updates mentioned.",{"category":144,"check":151,"severity":85,"summary":152},"Error Handling","The README and architecture notes mention error events and corrupt-line tolerance, but the specifics of error catching, categorization, and structured reporting are not detailed.",{"category":144,"check":154,"severity":85,"summary":155},"Logging","The extension mentions an audit log for destructive invocations and an event log, but details on structured audit records and their accessibility are not fully elaborated.",{"category":157,"check":158,"severity":63,"summary":159},"Compliance","GDPR","The extension operates on local project files and does not inherently handle or submit personal data to third parties.",{"category":157,"check":161,"severity":63,"summary":162},"Target market","The extension is designed for general AI coding agent use and does not exhibit any regional or jurisdictional logic.",{"category":131,"check":164,"severity":63,"summary":165},"Runtime stability","The extension relies on standard POSIX tools like bash and jq, and its installation script handles Windows via PowerShell/Git Bash, suggesting good cross-platform compatibility.",{"category":83,"check":167,"severity":63,"summary":168},"README","The README is comprehensive, clearly states the extension's purpose, and provides detailed information on its functionality and usage.",{"category":72,"check":170,"severity":63,"summary":171},"Tool surface size","The MCP server exposes 7 tools, which falls within the desired range of 3-10.",{"category":79,"check":173,"severity":63,"summary":174},"Overlapping near-synonym tools","The exposed MCP tools have distinct names and purposes, avoiding near-synonyms that could cause confusion.",{"category":83,"check":176,"severity":63,"summary":177},"Phantom features","All features mentioned in the README, such as Claude Code hooks, MCP server tools, and session continuity, appear to be implemented and supported by the code.",{"category":179,"check":180,"severity":63,"summary":181},"Install","Installation instruction","The README provides clear, copy-pasteable installation instructions for macOS/Linux/WSL and Windows, including necessary setup steps.",{"category":183,"check":184,"severity":85,"summary":185},"Errors","Actionable error messages","While the extension mentions error logging and tolerance, the details on specific error message content (what, why, remediation) are not explicitly provided in the documentation.",{"category":187,"check":188,"severity":63,"summary":189},"Execution","Pinned dependencies","The extension relies on standard system tools (`bash`, `jq`, `git`) and does not bundle external scripts with specific interpreter or side-effect headers.",{"category":72,"check":191,"severity":63,"summary":192},"Dry-run preview","The extension focuses on memory management and analysis, not destructive operations, thus a dry-run mode is not applicable or necessary.",{"category":194,"check":195,"severity":63,"summary":196},"Protocol","Idempotent retry & timeouts","The extension operates locally and synchronously for most operations, and does not involve remote calls that would require complex retry or timeout logic.",{"category":157,"check":198,"severity":63,"summary":199},"Telemetry opt-in","The extension does not appear to emit any telemetry, thus the opt-in/out status is not applicable.",{"category":72,"check":201,"severity":63,"summary":202},"Theme declaration","The README clearly defines a niche curation theme around 'git-native memory' for AI coding agents, emphasizing zero dependencies and local operation.",{"category":140,"check":204,"severity":63,"summary":205},"Per-entry version metadata","The `marketplace.json` entry includes an explicit `version` field ('1.2.0') alongside the source reference.",1778688414563,"This extension provides a persistent, git-native memory layer for AI coding agents like Claude Code and Codex, utilizing bash and jq scripts to track file operations and git history. It avoids external services, databases, or embeddings, offering a lightweight and auditable context memory that survives agent restarts and session clears.",[209,210,211,212,213,214,215],"Persistent, git-native memory layer","Zero external dependencies beyond jq","Works natively with Claude Code hooks","MCP server for Codex, Cursor, Windsurf","Session continuity with work snapshots","Co-change prediction for related modules","Local file operations and git history tracking",[217,218,219,220],"Providing a vector database or embedding store","Requiring external services or daemons","Replacing core AI agent functionality with opaque logic","Storing or processing sensitive user secrets",[222,223,224],"Evidence-based reasoning","Local-first architecture","Developer workflow augmentation",[32,33,226],"git (optional, recommended)","3.1.0","To enable AI coding agents to retain project context across sessions and interactions, leading to more efficient and accurate code generation by providing a persistent, evidence-based memory.","The extension has a strong technical foundation with clear documentation and a well-defined scope. The primary area for improvement is the lack of detailed error message reporting and specific configuration parameter documentation.",93,"A highly capable, self-contained memory layer for AI coding agents, leveraging local file operations and git history.",[233,28,234,235,236,32,33],"ai","code","developer-tools","git",[238,239,240,241],"Maintaining project context for AI coding agents across multiple sessions","Providing structured, auditable context to AI agents to ensure rule adherence","Enabling teams to share AI agent knowledge via shared git repositories","Reducing LLM token costs by performing heavy analysis locally",[243,244,245,246,247,248,249],"User interacts with AI agent (Claude Code / Codex)","Agent reads/writes files, triggering hooks","Hooks record events locally","Background analysis mines git history and events for patterns","Work snapshot saved at session end","Snapshot injected at session start, providing continuity","MCP server exposes knowledge graph data to other agents",{"codeQuality":251,"collectedAt":253,"documentation":254,"maintenance":257,"security":261,"testCoverage":262},{"hasLockfile":252},false,1778688382952,{"descriptionLength":255,"readmeSize":256},72,11113,{"closedIssues90d":11,"forks":11,"hasChangelog":258,"openIssues90d":41,"pushedAt":259,"stars":260},true,1778039913000,10,{"hasNpmPackage":252,"license":38,"smitheryVerified":252},{"hasCi":252,"hasTests":258},{"updatedAt":264},1778688415198,{"basePath":18,"githubOwner":19,"githubRepo":15,"locale":20,"slug":15,"type":266},"marketplace",null,{"evaluate":269,"extract":271},{"promptVersionExtension":227,"promptVersionScoring":25,"score":230,"tags":270,"targetMarket":34,"tier":35},[233,28,234,235,236,32,33],{"commitSha":37,"license":38,"marketplace":272,"plugin":274},{"name":273,"pluginCount":41},"hilyfux-knowledge-graph",{"mcpCount":11,"provider":40,"skillCount":11},{"repoId":43},{"_creationTime":277,"_id":43,"identity":278,"providers":279,"workflow":338},1778688377223.2065,{"githubOwner":19,"githubRepo":15,"sourceUrl":16},{"classify":280,"discover":332,"github":335},{"commitSha":37,"extensions":281},[282,295,306],{"basePath":18,"description":51,"displayName":273,"installMethods":283,"rationale":284,"selectedPaths":285,"source":294,"sourceLanguage":20,"type":266},{"claudeCode":53},"marketplace.json at .claude-plugin/marketplace.json",[286,289,291],{"path":287,"priority":288},".claude-plugin/marketplace.json","mandatory",{"path":290,"priority":288},"README.md",{"path":292,"priority":293},"LICENSE","high","rule",{"basePath":18,"description":13,"displayName":15,"installMethods":296,"license":38,"rationale":297,"selectedPaths":298,"source":294,"sourceLanguage":20,"type":21},{"claudeCode":15},"plugin manifest at .claude-plugin/plugin.json (coalesced with duplicate plugin at standalone/skills/knowledge-graph)",[299,301,302,303],{"path":300,"priority":288},".claude-plugin/plugin.json",{"path":290,"priority":288},{"path":292,"priority":293},{"path":304,"priority":305},"skills/knowledge-graph/SKILL.md","medium",{"basePath":307,"description":308,"displayName":15,"installMethods":309,"rationale":310,"selectedPaths":311,"source":294,"sourceLanguage":20,"type":331},"skills/knowledge-graph","Use when user says \"update/refresh knowledge graph\", \"graph status\", \"blind spots\", \"knowledge node coverage\", \"CLAUDE.md coverage\", or \"init knowledge graph\". Also use when receiving a \"[kg auto-trigger]\" message injected by hooks. Do not use for regular coding tasks.\n",{"claudeCode":53},"SKILL.md frontmatter at skills/knowledge-graph/SKILL.md (coalesced with duplicate skill at standalone/skills/knowledge-graph)",[312,314,317,319,321,323,325,327,329],{"path":313,"priority":288},"SKILL.md",{"path":315,"priority":316},"scripts/analyze.sh","low",{"path":318,"priority":316},"scripts/context.sh",{"path":320,"priority":316},"scripts/guard.sh",{"path":322,"priority":316},"scripts/infer.sh",{"path":324,"priority":316},"scripts/mcp-server.sh",{"path":326,"priority":316},"scripts/prompt-trigger.sh",{"path":328,"priority":316},"scripts/track.sh",{"path":330,"priority":316},"scripts/version.sh","skill",{"sources":333},[334],"manual",{"closedIssues90d":11,"description":336,"forks":11,"license":38,"openIssues90d":41,"pushedAt":259,"readmeSize":256,"stars":260,"topics":337},"Stop AI Coding from forgetting.  A knowledge graph–driven memory layer for LLMs (ChatGPT, Claude, Codex, DeepSeek, Gemini), enabling persistent long-term memory beyond context window limits.  Build smarter AI agents with structured context, better consistency, and scalable multi-step reasoning across complex coding workflows.",[],{"classifiedAt":339,"discoverAt":340,"extractAt":341,"githubAt":341,"updatedAt":339},1778688381018,1778688377223,1778688379278,[233,32,234,235,236,33,28],{"evaluatedAt":264,"extractAt":47,"updatedAt":344},1778688496391,[],[347,379,409,438,467],{"_creationTime":348,"_id":349,"community":350,"display":351,"identity":357,"providers":359,"relations":372,"tags":374,"workflow":375},1778697668802.9524,"k1746m1dnjtegrp9ftydmb5wn186mxsv",{"reviewCount":11},{"description":352,"installMethods":353,"name":355,"sourceUrl":356},"One-stop shop for building AI-powered products and businesses with Stripe.",{"claudeCode":354},"stripe/agent-toolkit","stripe","https://github.com/stripe/agent-toolkit",{"basePath":18,"githubOwner":355,"githubRepo":358,"locale":20,"slug":358,"type":266},"agent-toolkit",{"evaluate":360,"extract":369},{"promptVersionExtension":227,"promptVersionScoring":25,"score":361,"tags":362,"targetMarket":34,"tier":368},100,[355,233,363,364,365,366,367],"sdk","typescript","python","llm","agent","verified",{"commitSha":37,"marketplace":370,"plugin":371},{"name":355,"pluginCount":41},{"mcpCount":11,"provider":40,"skillCount":11},{"repoId":373},"kd74b9k48z93rf42bgdvfdcahh86nzpd",[367,233,366,365,363,355,364],{"evaluatedAt":376,"extractAt":377,"updatedAt":378},1778697682749,1778697668803,1778697814342,{"_creationTime":380,"_id":381,"community":382,"display":383,"identity":389,"providers":392,"relations":402,"tags":404,"workflow":405},1778688322101.8252,"k17944zm3ehfvm4ntncyz1dzyx86nc6v",{"reviewCount":11},{"description":384,"installMethods":385,"name":387,"sourceUrl":388},"Topic-based automatic memory for Claude Code",{"claudeCode":386},"hatawong/claude-recap","claude-recap-marketplace","https://github.com/hatawong/claude-recap",{"basePath":18,"githubOwner":390,"githubRepo":391,"locale":20,"slug":391,"type":266},"hatawong","claude-recap",{"evaluate":393,"extract":399},{"promptVersionExtension":227,"promptVersionScoring":25,"score":361,"tags":394,"targetMarket":34,"tier":368},[28,395,396,397,398],"context","cli","persistence","automation",{"commitSha":37,"marketplace":400,"plugin":401},{"name":387,"pluginCount":41},{"mcpCount":11,"provider":40,"skillCount":11},{"repoId":403},"kd78y3gm1ky53msejxede6b4x986nqyc",[398,396,395,28,397],{"evaluatedAt":406,"extractAt":407,"updatedAt":408},1778688340018,1778688322101,1778688561034,{"_creationTime":410,"_id":411,"community":412,"display":413,"identity":419,"providers":422,"relations":432,"tags":434,"workflow":435},1778683562157.875,"k174pnm5ch9ab6fr1etef2f2b586m74b",{"reviewCount":11},{"description":414,"installMethods":415,"name":417,"sourceUrl":418},"Persistent memory and cognitive profiling plugins for Claude Code",{"claudeCode":416},"cdeust/Cortex","cortex-plugins","https://github.com/cdeust/Cortex",{"basePath":18,"githubOwner":420,"githubRepo":421,"locale":20,"slug":421,"type":266},"cdeust","Cortex",{"evaluate":423,"extract":429},{"promptVersionExtension":227,"promptVersionScoring":25,"score":361,"tags":424,"targetMarket":34,"tier":368},[28,425,31,30,15,426,427,428],"cognitive-profiling","codebase-analysis","postgresql","pgvector",{"commitSha":37,"marketplace":430,"plugin":431},{"name":417,"pluginCount":41},{"mcpCount":11,"provider":40,"skillCount":11},{"repoId":433},"kd79gxpemvkr09a7zsb3h8kmah86nvgf",[30,426,425,15,31,28,428,427],{"evaluatedAt":436,"extractAt":437,"updatedAt":436},1778683583007,1778683562157,{"_creationTime":439,"_id":440,"community":441,"display":442,"identity":448,"providers":451,"relations":461,"tags":463,"workflow":464},1778698235845.4075,"k17c6qmv4dnjycsp8aa4wyfbgh86n3jd",{"reviewCount":11},{"description":443,"installMethods":444,"name":446,"sourceUrl":447},"Context7 Platform -- Up-to-date code documentation for LLMs and AI code editors",{"claudeCode":445},"upstash/context7","Context7 Platform","https://github.com/upstash/context7",{"basePath":18,"githubOwner":449,"githubRepo":450,"locale":20,"slug":450,"type":266},"upstash","context7",{"evaluate":452,"extract":457},{"promptVersionExtension":227,"promptVersionScoring":25,"score":453,"tags":454,"targetMarket":34,"tier":368},98,[455,234,366,233,235,456],"documentation","lsp",{"commitSha":37,"license":38,"marketplace":458,"plugin":460},{"name":459,"pluginCount":41},"context7-marketplace",{"mcpCount":11,"provider":40,"skillCount":11},{"repoId":462},"kd7955sg5wbf89gw527wdep66n86na9w",[233,234,235,455,366,456],{"evaluatedAt":465,"extractAt":466,"updatedAt":465},1778698251135,1778698235845,{"_creationTime":468,"_id":469,"community":470,"display":471,"identity":477,"providers":480,"relations":488,"tags":490,"workflow":491},1778693480094.2803,"k176y9t6x729cqdcsy39zh9agh86mbr3",{"reviewCount":11},{"description":472,"installMethods":473,"name":475,"sourceUrl":476},"A calm, CLI-native way to semantically grep everything, like code, images, pdfs and more.",{"claudeCode":474},"mixedbread-ai/mgrep","Mixedbread-Grep","https://github.com/mixedbread-ai/mgrep",{"basePath":18,"githubOwner":478,"githubRepo":479,"locale":20,"slug":479,"type":266},"mixedbread-ai","mgrep",{"evaluate":481,"extract":485},{"promptVersionExtension":227,"promptVersionScoring":25,"score":26,"tags":482,"targetMarket":34,"tier":35},[483,396,234,484,366],"search","indexing",{"commitSha":37,"marketplace":486,"plugin":487},{"name":475,"pluginCount":41},{"mcpCount":11,"provider":40,"skillCount":11},{"repoId":489},"kd7713fq0jzdyq6gzkd3q3t4f586ngft",[396,234,484,366,483],{"evaluatedAt":492,"extractAt":493,"updatedAt":494},1778693495991,1778693480094,1778693603013]