Awesome AI Agents › End-to-End Optimizers
Headroom is an innovative solution designed to optimize the efficiency of AI agents and Large Language Models (LLMs) by significantly reducing the token count of input data. It achieves 60-95% token savings on various data types, including tool outputs, logs, RAG chunks, files, and conversational history, without compromising the accuracy of the LLM's responses. The project offers multiple integration methods: as a Python or TypeScript library for inline compression, as a proxy for seamless integration with existing tools (supporting languages like Claude Code, Cursor, and LangChain), and as an MCP (Multi-Compotent Protocol) server with functions like `headroom_compress` and `headroom_retrieve`. Key features include a `ContentRouter` that intelligently selects the appropriate compression algorithm (e.g., `SmartCrusher` for JSON, `CodeCompressor` for AST, and `Kompress-base` for prose), an `CacheAligner` for optimizing provider KV cache hits, and a Reversible Compression (CCR) system that caches original content locally, allowing LLMs to retrieve it on demand. Headroom also incorporates cross-agent memory for shared storage across different LLMs and agents, `headroom learn` for mining failed sessions to write corrections, and an output token reduction feature that trims unnecessary ceremony or redundant
https://github.com/headroomlabs-ai/headroom
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 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.
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.
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.
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.
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.
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.
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.