swe-agent

Democratizing Reinforcement Learning for LLMs

Sandboxed code execution for AI agents, locally or on the cloud. Massively parallel, easy to extend. Powering SWE-agent and more.

Simple to install, powerful command-line based AI agent system for coding.

RepoMaster: The open-source AI agent that masters GitHub. It turns any code repository into a powerful tool, achieving a new level of autonomous task-solving. An open alternative to Claude-Code.

This repo tracks the opened and merged PRs by the top SWE coding agents by OpenAI, GitHub, and others. Updates regularly.

SE-Agent is a self-evolution framework for LLM Code agents. It enables trajectory-level evolution to exchange information across reasoning paths via Revision, Recombination, and Refinement, expanding the search space and escaping local optima. On SWE-bench Verified, it achieves SOTA performance

Repo-level benchmark for real-world Code Agents: from repo understanding → env setup → incremental dev/bug-fixing → task delivery, with cost-aware α metric.

Container-free RL framework for training software engineering agents

We track and analyze the activity and performance of autonomous code agents in the wild

Agentic Harness for the LLVM Compiler

Your AI-powered SWE teammate, built into your git workflow

An Always-On Autonomous AI Software Engineering Team built with LangGraph, FastAPI, and Next.js that operates natively within your GitHub repositories.

An autonomous AI coding agent with novel innovations in tool state management and code editing, running in a Docker sandbox with a persistent shell, parallel tools, and a VNC desktop.

Autonomous Software Engineering Agents — self-healing, self-diagnosing development team powered by Claude Code and A2A protocol

Autonomous AI coding agent triggered by Linear tickets — spins up full dev environments on EC2, writes code, runs tests, and opens PRs with live preview URLs.

[ACL 2026 Findings] SWE-AGILE: A Software Agent Framework for Efficiently Managing Dynamic Reasoning Context

A curated list of training & evaluation environments for LLM/VLM agents (SWE-Gym, GEM, RAGEN, AgentGym, WebArena, OSWorld, ToolBench…). Updated weekly.

Lawful Good Rust Agentic Operating System — production-grade agent harness with WASM sandboxing, heterogeneous inference, SurrealDB memory, sentinel guardrails.

cli-first harness agnostic agent orchestration tool

Drive Cursor and any ACP-compatible agent from your terminal — with persistent sessions and full automation support.