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
lupantech/AgentFlow
AgentFlow is a trainable, tool-integrated agentic framework from the paper on in-the-flow agentic system optimization, accepted at ICLR 2026. Instead of training one language model to interleave reasoning with tool calls, it splits the system into four specialised modules that coordinate over multiple turns through an evolving memory: a Planner that decides the next step, an Executor that runs tools, a Verifier that checks intermediate results, and a Generator that produces the final answer. Tools available to the Executor include a base generator, a Python coder, Google search, Wikipedia search and general web search, and the tool set can be extended. The distinguishing contribution is the training method: Flow-based Group Refined Policy Optimization, or Flow-GRPO, which optimises the Planner in place inside the running system and online, presented as a way to handle long-horizon reasoning with sparse rewards where credit for the final outcome must be assigned back to individual planning decisions. Reported results show a seven billion parameter backbone beating stronger baselines on ten benchmarks, with gains of roughly fifteen percent on search, fourteen percent on agentic tasks, fifteen percent on mathematics and four percent on science. The repository provides a setup script that builds a virtual environment, an environment template for API keys covering OpenAI for judging responses, Google for the search tool and DashScope or Together for calling Qwen2.5-7B-Instruct as the engine behind agents and tools, plus instructions for serving that model locally with vLLM. It ships quick start paths for inference and for Flow-GRPO training, dataset preparation steps, a benchmark suite that can be run in parallel, and guidance for substituting a custom model. A Gradio demo, released model weights and a tutorial video are linked from the README.
https://github.com/lupantech/AgentFlow
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.