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
Gen-Verse/Open-AgentRL
Open-AgentRL from Gen-Verse is an open-source reinforcement learning codebase for training large language models in agentic settings, packaging several research projects into one repository. It supports the common policy optimisation algorithms used for language models, including GRPO, PPO and RLHF-style training, and targets multi-turn rollouts where the model reasons, calls tools and receives feedback across many steps rather than producing a single answer. Distributed training is supported for larger runs. The headline component is RLAnything, an ICML 2026 paper whose framework treats the environment, the policy and the reward model as parts of one closed loop instead of fixed inputs. The policy is trained on a combination of outcome signals and step-wise signals produced by a reward model; the reward model is itself optimised through consistency feedback, so improvements in one feed back into the other; and a theory-motivated automatic environment adaptation mechanism adjusts the training environment using critic feedback from both models, which the authors describe as learning from experience. The paper reports ablations showing each added component improving the overall system, and argues that step-wise signals from an optimised reward model beat outcome signals derived from human labels. The repository also carries AutoTool, a second ICML 2026 project concerned with tool use, and DemyAgent, a released agent model, with trained policy and reward checkpoints published as a Hugging Face collection alongside the code. Coverage extends to coding agents and GUI agents as training scenarios, and the tags list entropy-based methods and reasoning-oriented training. The audience is researchers and engineers who want to reproduce or extend agentic RL results rather than deploy an agent in production, so the material is organised around papers, configurations and released checkpoints.
https://github.com/Gen-Verse/Open-AgentRL
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.