[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80708":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":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":14,"stars90d":16,"forks30d":16,"starsTrendScore":19,"compositeScore":20,"rankGlobal":10,"rankLanguage":10,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":22,"hasPages":24,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":46,"readmeContent":47,"aiSummary":48,"trendingCount":16,"starSnapshotCount":16,"syncStatus":13,"lastSyncTime":49,"discoverSource":50},80708,"kyros-ai","Kyros-494\u002Fkyros-ai","Kyros-494","Kyros — The Memory OS for AI Agents  Give your AI agents secure, self-correcting, persistent memory in 3 lines of code.  Three memory types (episodic, semantic, procedural) with built-in forgetting curves, cryptographic integrity, and automatic contradiction resolution. Model-agnostic REST API with Python and TypeScript SDKs.","",null,"Python",88,2,44,1,0,15,26,45,78.83,"Apache License 2.0",false,"master",true,[26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45],"agents","ai","artificial-intelligence","docker","episodic-memory","fastapi","langchain","llamaindex","llm","memory","open-source","postgresql","procedural-memory","python","rag","redis","retrieval-augmented-generation","semantic-memory","typescript","vector-database","2026-06-12 04:01:29","\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002FKyros-494\u002Fkyros-ai\u002Fmaster\u002Fdocs\u002Fassets\u002Fkyros-logo.png\" alt=\"Kyros Logo\" width=\"120\" \u002F>\n\u003C\u002Fdiv>\n\n\u003Ch1 align=\"center\">Kyros\u003C\u002Fh1>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>Persistent Memory System for AI Agents\u003C\u002Fstrong>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  Give your AI agents secure, self-correcting, persistent memory with cryptographic integrity and natural forgetting curves.\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fkyros-sdk\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fkyros-sdk?color=blue&label=PyPI\" alt=\"PyPI version\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@kyros\u002Fsdk\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002F@kyros\u002Fsdk?color=blue&label=npm\" alt=\"npm version\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Factions\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Factions\u002Fworkflow\u002Fstatus\u002FKyros-494\u002Fkyros-ai\u002Fci.yml?branch=main&label=CI\" alt=\"CI Status\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Fblob\u002Fmain\u002FLICENSE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache%202.0-blue.svg\" alt=\"License\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Fstargazers\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FKyros-494\u002Fkyros-ai?style=social\" alt=\"GitHub Stars\" \u002F>\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#features\">Features\u003C\u002Fa> •\n  \u003Ca href=\"#quick-start\">Quick Start\u003C\u002Fa> •\n  \u003Ca href=\"#documentation\">Documentation\u003C\u002Fa> •\n  \u003Ca href=\"#examples\">Examples\u003C\u002Fa> •\n  \u003Ca href=\"#contributing\">Contributing\u003C\u002Fa> •\n  \u003Ca href=\"#license\">License\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n## Overview\n\nKyros is an open-source memory operating system designed specifically for AI agents. It provides three biologically-inspired memory types (episodic, semantic, and procedural) with built-in features like cryptographic integrity proofs, Ebbinghaus decay curves, and belief propagation.\n\n### The Problem\n\nModern AI agents are **stateless** — they forget everything between sessions. Developers manually implement memory using:\n- Vector databases for semantic search\n- Key-value stores for facts\n- Custom logic for memory management\n- Ad-hoc solutions for data integrity\n\nThis approach is **fragile, insecure, and doesn't scale**.\n\n### The Solution\n\nKyros provides a **complete memory system** in a single SDK:\n\n```python\nimport kyros\n\nclient = kyros.Client(api_key=\"your_api_key\")\n\n# Store a memory\nclient.remember(\"agent-1\", \"User prefers Python for backend development\")\n\n# Recall relevant memories\nresults = client.recall(\"agent-1\", \"What language does the user prefer?\")\nprint(results.results[0].content)\n# → \"User prefers Python for backend development\"\n```\n\n---\n\n## Features\n\n### Three Memory Types\n\nBased on cognitive science research, Kyros implements three distinct memory systems:\n\n- **Episodic Memory**: Events and experiences (conversations, actions, observations)\n- **Semantic Memory**: Facts and knowledge (user preferences, domain knowledge)\n- **Procedural Memory**: Skills and workflows (how to perform tasks)\n\n### Cryptographic Integrity\n\nEvery memory is protected with SHA-256 hashing and Merkle tree proofs:\n- Detect tampering and poisoning attacks\n- Verify memory authenticity\n- Audit trail for all changes\n- Immutable memory history\n\n### Ebbinghaus Decay\n\nMemories naturally fade over time based on category-specific decay rates:\n- Market data: 1.4 days half-life\n- User identity: 693 days half-life\n- Prevents memory bloat\n- Keeps context relevant\n\n### Belief Propagation\n\nWhen facts conflict, confidence updates ripple through the semantic graph:\n- Automatic conflict resolution\n- Confidence scoring\n- Relationship tracking\n- Self-correcting knowledge base\n\n### High Performance\n\n- **\u003C20ms** average recall latency\n- **\u003C50ms** memory storage\n- **pgvector** for semantic search\n- **Redis** caching layer\n- Horizontal scaling support\n\n### Framework Integrations\n\nWorks seamlessly with popular AI frameworks:\n- **LangChain**: Memory integration\n- **LlamaIndex**: Data connector\n- **AutoGen**: Multi-agent memory\n- **CrewAI**: Crew memory\n- **Vercel AI SDK**: Streaming support\n\n### Model Agnostic\n\nCompatible with any LLM provider:\n- OpenAI (GPT-4, GPT-3.5)\n- Anthropic (Claude)\n- Google (Gemini)\n- Open source models (Llama, Mistral)\n- Custom models\n\n---\n\n## Quick Start\n\n### Prerequisites\n\n- **Docker** and **Docker Compose** (for self-hosting)\n- **Python 3.11+** or **Node.js 18+** (for SDK)\n\n### Installation\n\n#### Option 1: Self-Hosted (Recommended)\n\n```bash\n# Clone the repository\ngit clone https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai.git\ncd kyros-ai\n\n# Start the server\ndocker compose up -d\n\n# Server runs at http:\u002F\u002Flocalhost:8000\n```\n\n#### Option 2: Cloud Hosted\n\nComing soon!!!\n\n### SDK Installation\n\n**Python:**\n```bash\npip install kyros-sdk\n```\n\n**TypeScript:**\n```bash\nnpm install @kyros\u002Fsdk\n```\n\n### Basic Usage\n\n**Python:**\n```python\nimport kyros\n\n# Initialize client\nclient = kyros.Client(\n    base_url=\"http:\u002F\u002Flocalhost:8000\",  # or https:\u002F\u002Fapi.kyros.ai\n    api_key=\"your_api_key\"\n)\n\n# Store episodic memory (what happened)\nclient.store_episode(\n    agent_id=\"agent-1\",\n    content=\"User asked about pricing plans\",\n    metadata={\"timestamp\": \"2026-05-03T10:30:00Z\"}\n)\n\n# Store semantic memory (what is true)\nclient.store_fact(\n    agent_id=\"agent-1\",\n    subject=\"user_123\",\n    predicate=\"subscription_plan\",\n    value=\"Pro\"\n)\n\n# Store procedural memory (how to do things)\nclient.store_procedure(\n    agent_id=\"agent-1\",\n    name=\"handle_refund_request\",\n    steps=[\"Verify purchase\", \"Check refund policy\", \"Process refund\"]\n)\n\n# Query memories\nresults = client.query(\n    agent_id=\"agent-1\",\n    query=\"What plan is the user on?\",\n    memory_types=[\"semantic\"]\n)\n\nfor memory in results.results:\n    print(f\"{memory.content} (confidence: {memory.confidence})\")\n```\n\n**TypeScript:**\n```typescript\nimport { KyrosClient } from '@kyros\u002Fsdk';\n\n\u002F\u002F Initialize client\nconst client = new KyrosClient({\n  baseUrl: 'http:\u002F\u002Flocalhost:8000',  \u002F\u002F or https:\u002F\u002Fapi.kyros.ai\n  apiKey: 'your_api_key'\n});\n\n\u002F\u002F Store and query memories\nawait client.storeEpisode({\n  agentId: 'agent-1',\n  content: 'User asked about pricing plans',\n  metadata: { timestamp: '2026-05-03T10:30:00Z' }\n});\n\nconst results = await client.query({\n  agentId: 'agent-1',\n  query: 'What plan is the user on?',\n  memoryTypes: ['semantic']\n});\n\nresults.results.forEach(memory => {\n  console.log(`${memory.content} (confidence: ${memory.confidence})`);\n});\n```\n\n---\n\n## Examples\n\n### LangChain Integration\n\n```python\nfrom langchain.memory import KyrosMemory\nfrom langchain.chains import ConversationChain\nfrom langchain.llms import OpenAI\n\n# Initialize Kyros memory\nmemory = KyrosMemory(\n    agent_id=\"chatbot-1\",\n    api_key=\"your_api_key\"\n)\n\n# Create conversation chain\nconversation = ConversationChain(\n    llm=OpenAI(),\n    memory=memory\n)\n\n# Chat with persistent memory\nresponse = conversation.predict(input=\"My name is Alice\")\n# Memory automatically stored\n\nresponse = conversation.predict(input=\"What's my name?\")\n# → \"Your name is Alice\"\n```\n\n### Multi-Agent System\n\n```python\nimport kyros\n\nclient = kyros.Client(api_key=\"your_api_key\")\n\n# Agent 1: Research agent\nclient.remember(\"researcher\", \"Found 3 relevant papers on neural networks\")\n\n# Agent 2: Writer agent queries researcher's findings\nresults = client.recall(\"researcher\", \"What papers did you find?\")\npapers = results.results[0].content\n\n# Writer uses the information\nclient.remember(\"writer\", f\"Writing article based on: {papers}\")\n```\n\n### Causal Reasoning\n\n```python\n# Store causal relationships\nclient.store_causal_edge(\n    agent_id=\"agent-1\",\n    cause_memory_id=\"mem_abc123\",\n    effect_memory_id=\"mem_def456\",\n    confidence=0.95\n)\n\n# Query causal chains\nchain = client.get_causal_chain(\n    agent_id=\"agent-1\",\n    memory_id=\"mem_def456\"\n)\n\n# Understand why something happened\nfor link in chain:\n    print(f\"{link.cause} → {link.effect} (confidence: {link.confidence})\")\n```\n\n---\n\n## Architecture\n\n```\n┌─────────────────────────────────────────────────────────────┐\n│                        Your Application                      │\n└─────────────────────────────────────────────────────────────┘\n                              │\n                              ▼\n┌─────────────────────────────────────────────────────────────┐\n│                      Kyros SDK (Python\u002FTS)                   │\n└─────────────────────────────────────────────────────────────┘\n                              │\n                              ▼\n┌─────────────────────────────────────────────────────────────┐\n│                    Kyros API Server (FastAPI)                │\n│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐      │\n│  │   Episodic   │  │   Semantic   │  │  Procedural  │      │\n│  │    Memory    │  │    Memory    │  │    Memory    │      │\n│  └──────────────┘  └──────────────┘  └──────────────┘      │\n│                                                               │\n│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐      │\n│  │   Integrity  │  │    Decay     │  │    Belief    │      │\n│  │    Proofs    │  │    Engine    │  │ Propagation  │      │\n│  └──────────────┘  └──────────────┘  └──────────────┘      │\n└─────────────────────────────────────────────────────────────┘\n                              │\n                              ▼\n┌─────────────────────────────────────────────────────────────┐\n│              Storage Layer (PostgreSQL + Redis)              │\n│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐      │\n│  │  PostgreSQL  │  │    Redis     │  │   pgvector   │      │\n│  │  (Memories)  │  │   (Cache)    │  │ (Embeddings) │      │\n│  └──────────────┘  └──────────────┘  └──────────────┘      │\n└─────────────────────────────────────────────────────────────┘\n```\n\n---\n\n## Documentation\n\n### Getting Started\n- [Introduction](.\u002Fdocs\u002Fintroduction.md) - Learn about Kyros and its architecture\n- [Quick Start Guide](.\u002Fdocs\u002Fquickstart.md) - Get up and running in 5 minutes\n- [Core Concepts](.\u002Fdocs\u002Fconcepts.md) - Understand memory types and features\n\n### SDK References\n- [Python SDK](.\u002Fdocs\u002Fpython-sdk.md) - Complete Python API reference\n- [TypeScript SDK](.\u002Fdocs\u002Ftypescript-sdk.md) - Complete TypeScript API reference\n\n### Deployment\n- [Self-Hosting Guide](.\u002Fdocs\u002Fself-hosting.md) - Deploy Kyros on your infrastructure\n- [Configuration](.\u002Fdocs\u002Fconfiguration.md) - Environment variables and settings\n- [Security Best Practices](.\u002FSECURITY.md) - Secure your deployment\n\n### Advanced Topics\n- [Architecture Deep Dive](.\u002Fdocs\u002Farchitecture.md) - System design and internals\n\n---\n\n\n## Community\n\n### Get Help\n\n- **Documentation**: [docs.kyros.ai](https:\u002F\u002Fdocs.kyros.ai)\n- **GitHub Discussions**: [Ask questions](https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Fdiscussions)\n- **GitHub Issues**: [Report bugs](https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Fissues)\n- **Email**: [kyros.494@gmail.com](mailto:kyros.494@gmail.com)\n\n### Stay Updated\n\n- **Star this repo** to follow development\n- **Watch releases** for new versions\n\n### Contributing\n\nWe welcome contributions from the community! See our [Contributing Guide](.\u002FCONTRIBUTING.md) for:\n- Code of Conduct\n- Development setup\n- Commit conventions\n- Pull request process\n- Testing requirements\n\n**Good First Issues**: [View beginner-friendly tasks →](https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Flabels\u002Fgood%20first%20issue)\n\n---\n\n## License\n\nKyros uses an **Open Core** licensing model:\n\n- **Server Core**: [Apache 2.0 License](.\u002FLICENSE) - Free to self-host and modify\n- **SDKs & Integrations**: [MIT License](.\u002FLICENSE-MIT) - Free to use in any project\n- **Enterprise Modules**: Commercial License - [Contact us](mailto:kyros.494@gmail.com)\n\nThis means you can:\n- ✅ Use Kyros for free in commercial projects\n- ✅ Self-host on your own infrastructure\n- ✅ Modify the source code\n- ✅ Distribute your modifications\n- ❌ Use the \"Kyros\" trademark without permission\n\nSee [LICENSE](.\u002FLICENSE) and [LICENSE-MIT](.\u002FLICENSE-MIT) for full terms.\n\n---\n\n## Security\n\nWe take security seriously. If you discover a security vulnerability, please email [kyros.494@gmail.com](mailto:kyros.494@gmail.com) instead of using the issue tracker.\n\n- **Response Time**: Within 48 hours\n- **Patch Timeline**: 7 days for critical issues\n- **Disclosure**: Coordinated disclosure policy\n\nSee [SECURITY.md](.\u002FSECURITY.md) for our full security policy.\n\n---\n\n## Acknowledgments\n\nKyros is built on the shoulders of giants:\n\n- **FastAPI** - Modern Python web framework\n- **PostgreSQL** - Reliable database\n- **pgvector** - Vector similarity search\n- **Redis** - High-performance caching\n- **Sentence Transformers** - Embedding models\n- **LangChain** - LLM application framework\n\nSpecial thanks to all our [contributors](https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\u002Fgraphs\u002Fcontributors) who make this project possible.\n\n---\n\n\u003Cdiv align=\"center\">\n  \u003Cp>\n    \u003Cstrong>Built with ❤️ by the Kyros team\u003C\u002Fstrong>\n  \u003C\u002Fp>\n  \u003Cp>\n    \u003Ca href=\"https:\u002F\u002Fkyros.ai\">Website\u003C\u002Fa> •\n    \u003Ca href=\"https:\u002F\u002Fdocs.kyros.ai\">Documentation\u003C\u002Fa> •\n    \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FKyros-494\u002Fkyros-ai\">GitHub\u003C\u002Fa> •\n  \u003C\u002Fp>\n  \u003Cp>\n    \u003Csub>Copyright © 2026 Kyros. All rights reserved.\u003C\u002Fsub>\n  \u003C\u002Fp>\n\u003C\u002Fdiv>\n","Kyros 是一个专为AI代理设计的持久内存系统，旨在通过三行代码赋予AI代理安全、自校正且持久的记忆功能。它支持三种记忆类型：情景记忆、语义记忆和程序记忆，并内置遗忘曲线、加密完整性以及自动矛盾解决机制。项目提供了模型无关的REST API接口及Python和TypeScript的SDK，便于开发者集成使用。Kyros适用于需要长期保持用户偏好、对话历史或特定任务执行流程等信息的AI应用中，如客户服务聊天机器人、个性化推荐引擎等领域。","2026-06-11 04:01:43","CREATED_QUERY"]