[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80384":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":15,"stars30d":15,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":16,"rankGlobal":10,"rankLanguage":10,"license":17,"archived":18,"fork":18,"defaultBranch":19,"hasWiki":18,"hasPages":18,"topics":20,"createdAt":10,"pushedAt":10,"updatedAt":33,"readmeContent":34,"aiSummary":35,"trendingCount":15,"starSnapshotCount":15,"syncStatus":13,"lastSyncTime":36,"discoverSource":37},80384,"mnemo","omermaksutii\u002Fmnemo","omermaksutii","Persistent memory for Claude Code. Your AI never starts from scratch again.","",null,"TypeScript",56,2,1,0,1.43,"MIT License",false,"main",[21,22,23,24,25,26,27,28,29,30,31,32],"ai","anthropic","claude-code","embeddings","hnsw","mcp","mcp-server","memory","onnx","persistent-memory","semantic-search","typescript","2026-06-12 02:04:01","# Mnemo\n\n> Persistent memory for Claude Code. Your AI never starts from scratch again.\n\n![demo](demo.gif)\n\n[![CI](https:\u002F\u002Fgithub.com\u002Fomermaksutii\u002Fmnemo\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg)](https:\u002F\u002Fgithub.com\u002Fomermaksutii\u002Fmnemo\u002Factions)\n[![npm: cli](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002F@mnemo-mcp\u002Fcli?label=%40mnemo-mcp%2Fcli&color=cb3837)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@mnemo-mcp\u002Fcli)\n[![npm: server](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002F@mnemo-mcp\u002Fserver?label=%40mnemo-mcp%2Fserver&color=cb3837)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@mnemo-mcp\u002Fserver)\n[![npm: core](https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002F@mnemo-mcp\u002Fcore?label=%40mnemo-mcp%2Fcore&color=cb3837)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@mnemo-mcp\u002Fcore)\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-blue.svg)](LICENSE)\n\nMnemo gives Claude Code a brain that survives across sessions. It captures decisions, conventions, and preferences — and recalls them by meaning, on demand, with sub-100ms semantic search.\n\n---\n\n## Install\n\n```bash\n# install the CLI\nnpm install -g @mnemo-mcp\u002Fcli\n\n# install Mnemo into Claude Code (skill + MCP server)\nmnemo init\n\n# (optional) also wire auto-capture hooks\nmnemo init --with-hooks\n\n# verify\nmnemo doctor\n```\n\nAfter `mnemo init`, restart Claude Code. The `\u002Fmnemo` skill and the `mnemo_*` MCP tools are now available — Claude will use them automatically when relevant.\n\n## Use it from the terminal too\n\n```bash\nmnemo remember \"our API auth uses OAuth2 with refresh tokens every 30min\"\nmnemo remember --global \"I prefer pnpm over npm\"\nmnemo recall \"what's our auth pattern?\"\nmnemo list\nmnemo stats\nmnemo forget \u003Cid>\nmnemo export --out memories.json\nmnemo import memories.json\n```\n\n## What Mnemo solves\n\nClaude Code forgets everything when the session ends. `CLAUDE.md` partially helps — it's a static blob loaded on every turn — but it grows unwieldy fast and can't answer \"do we have a precedent for X?\"\n\nMnemo gives you semantic, on-demand memory:\n\n| | `CLAUDE.md` | Mnemo |\n|---|---|---|\n| Capacity | A handful of paragraphs before token cost hurts | Tens of thousands of memories |\n| Retrieval | Always loaded, every turn | On demand, by meaning |\n| Updates | You edit a file by hand | Captured automatically or via `\u002Fteach` |\n| Cross-project | Per-project only | Project + global tiers |\n| Forgetting | Manually delete lines | `mnemo forget \u003Cid>` or `\u002Fforget` |\n\n## How it works\n\n```\nClaude Code session\n  ├── \u002Fmnemo skill              ← teaches Claude when to call the tools\n  ├── @mnemo-mcp\u002Fserver server          ← exposes recall\u002Fremember\u002Fforget\u002Flist\u002Fstats\n  └── @mnemo-mcp\u002Fcli hooks           ← session-start, pre-task, post-edit auto-wiring\n              │\n              ▼\n       @mnemo-mcp\u002Fcore\n       ├── ONNX all-MiniLM-L6-v2  (384-dim embeddings, ~25MB, lazy)\n       ├── HNSW vector index      (sub-100ms recall at 50k memories)\n       └── sql.js (WASM SQLite)   (~\u002F.mnemo\u002Fmemory.db)\n```\n\nEverything is local-first. No daemon. No cloud account. No telemetry.\n\n## Auto-capture rules\n\nBy default the post-edit hook captures any change to:\n\n- `CLAUDE.md`, `AGENTS.md`, `GEMINI.md`\n- `*.adr.md`\n- `docs\u002Fdecisions\u002F**`\n- `docs\u002Fadr\u002F**`\n\nAnything you write to those files becomes a project memory automatically. Disable with `mnemo init` (no `--with-hooks`).\n\n## Recall scoring\n\n```\nscore = 0.7 × cosine_similarity + 0.2 × recency + 0.1 × access_boost\n```\n\nRecency decays with a 30-day half-life. Access boost saturates at 20 reads.\n\n## Layout\n\n```\n@mnemo-mcp\u002Fcore   — pure TS memory engine\n@mnemo-mcp\u002Fserver    — MCP server (drop into Claude Code)\n@mnemo-mcp\u002Fcli    — terminal commands + hook handlers + init\n```\n\n## Roadmap\n\n- ✅ **v0.1** — core engine + CLI\n- ✅ **v0.2** — MCP server (5 tools)\n- ✅ **v0.3** — hooks (session-start, pre-task, post-edit)\n- ✅ **v0.4** — `\u002Fmnemo` skill\n- ✅ **v1.0** — `mnemo init`, beefier doctor, CI, polish\n- ✅ **v1.1** — team mode, channels, secret guard, ingest, why, digest, dead, cite, interactive recall, JSON output, completions\n- ✅ **v1.2** — JSON mode in all commands, idempotent init, doctor robustness\n- ✅ **v2.0** — procedural memory: `mnemo procedure record\u002Frun\u002Fsuggest\u002Fdone` + 2 new MCP tools + auto-suggestion via `pre-task` hook\n- ⏳ **v2.1** — knowledge graph (entities + relations on top of vector layer)\n- ⏳ **v2.2** — self-reflective learning (auto-capture anti-patterns from `Stop`\u002F`SubagentStop` hooks)\n- ⏳ **v2.3** — cross-agent memory (Cursor, Aider, custom MCP)\n- ⏳ **v2.4** — plug-in framework (custom embedders, rankers, capture rules)\n- ⏳ **v2.5** — hosted optional sync + web UI\n\nSee [ROADMAP.md](ROADMAP.md) for design notes on each.\n\n## Development\n\n```bash\nnpm install\nnpm test                                    # fast (44 tests, hash embedder)\nMNEMO_TEST_ONNX=1 MNEMO_E2E=1 npm test     # full incl. real ONNX (~5s)\nnpm run build\nnpm run lint\n```\n\n## License\n\nMIT — see [LICENSE](LICENSE).\n","Mnemo 是一个为 Claude Code 提供持久化记忆的工具，确保 AI 不再从零开始。它通过捕捉决策、惯例和偏好，并利用语义搜索在需要时快速（小于100毫秒）召回这些信息，核心功能包括自动捕获、基于 ONNX 的嵌入模型以及 HNSW 向量索引技术。Mnemo 适合于任何希望增强 Claude Code 持久记忆能力的场景，无论是项目级还是全局级的知识管理都能得到支持。此外，该工具完全本地化运行，无需云服务或后台进程，保证了数据的安全与隐私。","2026-06-11 04:00:34","CREATED_QUERY"]