ИИ-агенты
memvid
A serverless, single-file memory layer for AI agents - replace complex RAG pipelines
★ 16,6 тыс. звёздApache-2.0
Смотреть руководство →Memary — проект с открытым исходным кодом из категории «ИИ-агенты» от kingjulio8238 (kingjulio8238/Memary на GitHub). У него 2 655 звёзд на GitHub и 205 форков, он распространяется по лицензии MIT и последний раз обновлялся на GitHub 22 октября 2024.
The open-source memory layer for autonomous agents — knowledge-graph memory that emulates human memory
Decide whether you are adding memory to your own agent, or evaluating Memary through its bundled personal-agent demo.
pip install memary (Python <= 3.11.9), or clone the repo and install requirements for the full demo.
Create a .env with your graph database credentials (free FalkorDB or Neo4j instance) and any API keys, choose local Ollama models or OpenAI models, and edit the user/system persona files.
Launch the Streamlit app, chat with the agent, and watch its knowledge graph and memory modules update in the dashboard.
Facts are stored in a graph database (FalkorDB or Neo4j) via LlamaIndex, with recursive retrieval and multi-hop reasoning over entity subgraphs.
After initialization, memory updates automatically as the agent interacts, and agent responses are written back into the graph.
Captures every entity inserted into the graph with timestamps, enabling timeline and theme analysis of a user's interests.
Tracks frequency and recency per entity to rank what the user knows deeply versus superficially.
A demo ChatAgent with default tools (search, vision via LLaVA, location via Google Maps, stocks), plus add_tool/remove_tool for custom tools.
With FalkorDB, multiple graphs keyed by user/agent ID let different agents keep separate memory contexts.
Clone Memary for the full demo; the library alone installs from PyPI in step 2.
git clone https://github.com/kingjulio8238/Memary.git
cd MemaryInstall Memary (Python <= 3.11.9 per the README) and, for a local checkout, its requirements.
pip install memary
# local checkout: pip install -r requirements.txtConfigure .env and the persona files first, then start the Streamlit app.
cd streamlit_app
streamlit run app.pykingjulio8238 created and maintains it; documentation is at kingjulio8238.github.io.
Yes, MIT-licensed. Its graph databases (FalkorDB/Neo4j) have free tiers, and models can run locally via Ollama.
Not necessarily — it defaults to local Ollama models. Hosted options (OpenAI, Perplexity, Google Maps, Alpha Vantage) use your own keys if you enable those tools.
Memobase focuses on user profiles for chatbots as a service. Memary is a research-oriented memory layer centered on a knowledge graph plus memory-stream/entity-store modules you embed yourself.
pip install memary, set up a free FalkorDB or Neo4j instance in .env, then run the Streamlit demo from the repo. Note the repo's last push was in October 2024.
A serverless, single-file memory layer for AI agents - replace complex RAG pipelines
Examples and integrations for Zep — the memory layer for AI agents.
A markdown knowledge graph with LSP for editors and MCP for agents.
M-flow is a bio-inspired cognitive memory engine for AI agents — a graph alternative to flat RAG
Real-time knowledge graphs that give agents a memory with structure.
Basic Memory: local-first MCP memory so you and your AI tools build a shared knowledge graph in Markdown
Другие проекты из категории «ИИ-агенты» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если Memary вам не подходит.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Соседние проекты из категории «ИИ-агенты» в каталоге, ближайшие по числу звёзд к Memary — продолжайте изучать полку по обе стороны от него.
Rust terminal multiplexer with a typed SDK: drive any CLI app from code.
Long-term memory runtime for AI agents — plain Markdown as truth, local ranked retrieval, no API key
NVIDIA's toolkit for connecting, profiling and optimizing teams of AI agents — across any framework
Multi-agent platform where every agent gets its own computer and long-term memory.
Bright Data's MCP server: web search, scraping, structured data and browser automation for agents (75 tools)
AI-native search that indexes code, docs, and PDFs with 40x fewer tokens than grep.
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.