[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80958":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":13,"subscribersCount":13,"size":13,"stars1d":13,"stars7d":13,"stars30d":13,"stars90d":13,"forks30d":13,"starsTrendScore":13,"compositeScore":13,"rankGlobal":10,"rankLanguage":10,"license":16,"archived":17,"fork":17,"defaultBranch":18,"hasWiki":17,"hasPages":17,"topics":19,"createdAt":10,"pushedAt":10,"updatedAt":20,"readmeContent":21,"aiSummary":22,"trendingCount":13,"starSnapshotCount":13,"syncStatus":15,"lastSyncTime":23,"discoverSource":24},80958,"memsy","memsy-io\u002Fmemsy","memsy-io","Monorepo for SDKs, Connectors and Docs","https:\u002F\u002Fdocs.memsy.io",null,"Python",30,0,31,2,"MIT License",false,"main",[],"2026-06-12 02:04:09","# Memsy\n\nPersistent memory for AI agents and applications. 88.12% on LoCoMo — the highest accuracy of any memory library at k≤20, at 4× better score-per-token than mem0.\n\n📚 **Docs**: [docs.memsy.io](https:\u002F\u002Fdocs.memsy.io)  |  🌐 **Website**: [memsy.io](https:\u002F\u002Fmemsy.io)\n\n---\n\n## SDKs\n\n| Language | Package | Install |\n|----------|---------|---------|\n| Python | [memsy-io\u002Fsdks\u002Fpython](https:\u002F\u002Fgithub.com\u002Fmemsy-io\u002Fmemsy\u002Ftree\u002Fmain\u002Fsdks\u002Fpython) | `pip install memsy` |\n| Node.js | [memsy-io\u002Fsdks\u002Fnode](https:\u002F\u002Fgithub.com\u002Fmemsy-io\u002Fmemsy\u002Ftree\u002Fmain\u002Fsdks\u002Fnode) | `npm install @memsy-io\u002Fmemsy` |\n\n## Quick Start\n\n**Python**\n```python\nfrom memsy import MemsyClient, EventPayload\n\nclient = MemsyClient(base_url=\"https:\u002F\u002Fapi.memsy.io\", api_key=\"msy_...\")\n\n# Store a memory\nclient.ingest([EventPayload(\n    actor_id=\"user_1\",\n    session_id=\"session_1\",\n    kind=\"user_message\",\n    content=\"I prefer dark mode in all apps\",\n)])\n\n# Recall it later\nresults = client.search(\"user preferences\", actor_id=\"user_1\")\nfor r in results.results:\n    print(r.content)\n```\n\n**Node.js**\n```ts\nimport { MemsyClient } from \"@memsy-io\u002Fmemsy\";\n\nconst client = new MemsyClient({ baseUrl: \"https:\u002F\u002Fapi.memsy.io\", apiKey: \"msy_...\" });\n\n\u002F\u002F Store a memory\nawait client.ingest([{\n  actorId: \"user_1\",\n  sessionId: \"session_1\",\n  kind: \"user_message\",\n  content: \"I prefer dark mode in all apps\",\n}]);\n\n\u002F\u002F Recall it later\nconst { results } = await client.search(\"user preferences\", { actorId: \"user_1\" });\nresults.forEach(r => console.log(r.content));\n```\n\n## Why Memsy\n\n| | Memsy | mem0 |\n|---|---|---|\n| LoCoMo Score | **88.12%** | 82.7% |\n| Tokens per query | ~1,700 | ~7,000 |\n| Score \u002F 1K tokens | **51.8** | 13.1 |\n\nBenchmark: full LoCoMo suite (1,540 questions), GPT-4.1 Mini, single pass, no post-processing. [Full results →](https:\u002F\u002Fgithub.com\u002Fmemsy-io\u002Fmemsy\u002Fblob\u002Fmain\u002Fbenchmark\u002FBENCHMARK_RESULTS.md)\n\n## Connectors\n\nComing soon: Claude MCP, OpenAI, LangChain, LlamaIndex.\n\n## Repository Structure\n\n```\nmemsy\u002F\n├── sdks\u002F\n│   ├── python\u002F       ← Python SDK (pip install memsy)\n│   └── node\u002F         ← Node.js SDK (npm install @memsy-io\u002Fmemsy)\n└── connectors\u002F       ← Claude, OpenAI and other connectors (coming soon)\n```\n\n## Contributing\n\nWe welcome contributions to any SDK or connector in this repo. See [CONTRIBUTING.md](CONTRIBUTING.md) for the full guide — repo layout, per-SDK setup, and PR conventions.\n\nFor bugs or feature requests, [open an issue](https:\u002F\u002Fgithub.com\u002Fmemsy-io\u002Fmemsy\u002Fissues).\n\n## License\n\nMIT\n","Memsy 是一个为AI代理和应用程序提供持久内存的项目。其核心功能包括高效存储与检索信息，支持Python和Node.js两种语言的SDK，并在LoCoMo基准测试中达到了88.12%的准确率，远超同类库。技术特点上，Memsy不仅提供了出色的查询效率（每千个令牌得分51.8），还计划集成Claude MCP、OpenAI等连接器以扩展其适用范围。此项目适用于需要长期记忆能力的各种AI应用开发场景，如聊天机器人、个性化推荐系统等。","2026-06-11 04:03:00","CREATED_QUERY"]