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
statelyai/agent
Stately Expert is a sophisticated framework designed for building AI agents powered by state machines, leveraging the XState library. Unlike traditional large language model (LLM) based agents that rely solely on text generation, Stately Expert enhances agent behavior by incorporating structured state machines to guide decision-making processes. This approach allows agents to utilize observations, message history, and feedback to make more informed and context-aware decisions. The framework supports custom planning capabilities, enabling agents to achieve specific goals through state machine logic combined with real-time observations and feedback. A key feature of Stately Expert is its integration with the Vercel AI SDK, which facilitates support for multiple AI model providers such as OpenAI, Anthropic, Google, Mistral, Groq, and Perplexity. This flexibility allows developers to choose or switch between different models seamlessly. The core concepts of the framework include observations, which track state transitions and metadata; decisions, which are actions taken based on current and goal states along with past experiences; feedback, which serves as rewards or evaluations to improve agent learning; insights, providing additional context to understand cause and effect; and episodes, representing complete sequences of state transitions akin to reinforcement learning episodes. Stately Expert aims to overcome limitations of traditional LLM agents by offering structured decision-making through state machines, enabling agents to learn from experience via feedback and observations, and providing contextual awareness for improved decision quality. The framework supports multiple decision-making policies and optional storage integration for persisting experiences and learning data. Overall, Stately Expert is a powerful tool for developers looking to create intelligent, adaptive AI agents that combine the rigor of state machines with the flexibility of modern AI models, enhancing both decision-making and learning capabilities.
https://github.com/statelyai/agent
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.
getzep/zep
Zep is an AI memory foundation that uses a temporal Knowledge Graph to enable continuous learning and personalized experiences for AI agents by efficiently managing and retrieving contextual chat and data artifacts.
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.