headroomlabs-ai/headroom
Headroom is a context compression layer for AI agents, reducing token usage by 60-95% across tool outputs, logs, RAG chunks, files, and conversation history, while maintaining accuracy.
Awesome AI Agents › End-to-End Optimizers
getzep/zep
Zep is an advanced AI memory foundation designed to enhance AI agents by enabling continuous learning from user interactions, thereby powering personalized experiences and improving accuracy. It operates by adding chat messages or data artifacts during user interactions or agent events, integrating this new information into a temporal Knowledge Graph that maintains contextual information about facts, including their validity over time. This allows AI agents to reason about state changes and track changes in user preferences, traits, or environments effectively. Zep is optimized for speed, precomputing facts, entity summaries, and other artifacts asynchronously to ensure fast retrieval, which primarily depends on the embedding service's performance. It supports various data types, including chat history messages and soon JSON and unstructured text, and manages chat session, user, and group-level graphs to capture organizational knowledge. Zep is framework-agnostic and can be integrated with popular AI frameworks like LangChain, LangGraph, Chainlit, and Microsoft Autogen. The project offers SDKs in Python, TypeScript, and Go, providing simple APIs for persisting chat history memory and advanced retrieval using BM25, semantic, and graph search techniques. Zep uses a unique temporal Knowledge Graph library called Graphiti, which autonomously builds and maintains Knowledge Graphs while handling changing relationships and preserving historical context. This approach addresses the limitations of traditional Knowledge Graphs that struggle with state changes. Zep Community Edition is the open-source distribution of Zep, sharing APIs with Zep Cloud, a managed service offering scalability, low latency, dialog classification, and structured data extraction. Zep is purpose-built for agent and assistant applications, supporting privacy compliance and efficient chat memory management. It is designed to overcome the challenges of using entire chat histories or traditional databases for memory, which often result in poor temporal reasoning and slow inference. Overall, Zep provides a robust, scalable, and efficient memory solution for AI applications requiring long-term memory and contextual understanding.
Also in End-to-End Optimizers
topoteretes/cognee
Cognee is an open-source project that provides scalable and modular memory solutions for AI agents using ECL pipelines, enabling efficient data ingestion, knowledge graph generation, and query capabilities to enhance AI performance and reduce hallucinations.
letta-ai/letta
Letta is an open-source framework for building stateful AI agents with advanced reasoning, transparent long-term memory, and context management, supporting multiple LLM backends and providing a graphical development environment.
langchain-ai/opengpts
OpenGPTs is an open-source platform that replicates and extends OpenAI's GPTs and Assistants API, offering customizable cognitive architectures, extensive tooling, and support for multiple language models and vector databases.
aiwaves-cn/agents
Agents 2.0 is an open-source framework that enables the training and self-evolution of autonomous language agents through symbolic learning inspired by neural network methodologies.
algorithmicsuperintelligence/optillm
OptiLLM is an optimizing inference proxy that significantly improves the accuracy and performance of large language models on reasoning tasks without requiring any model training, by applying over 20 advanced optimization techniques during inference.
mll-lab-nu/RAGEN
A reinforcement learning framework for training multi-turn reasoning LLM agents on the StarPO algorithm, shipping ten Gym-compatible environments plus diagnostics for echo-trap and reasoning-collapse failure modes.
langfengQ/verl-agent
verl-agent is an extension of veRL designed for scalable training of large language and vision-language model agents via reinforcement learning, featuring customizable input structures, diverse RL algorithms including GiGPO, and support for multi-modal environments.