AIエージェント
SWE-agent
Give it a GitHub issue and it tries to fix the code for you - the original SWE-bench agent
★ 20.5k スターMIT
ガイドを見る →SWE-benchは、SWE-benchが公開しているオープンソースのAIエージェントプロジェクトです（GitHub: SWE-bench/SWE-bench）。GitHubスターは5,986、フォークは1,004で、MITライセンスの下で公開されています。GitHubでの最終更新は2026年9月18日です。
The benchmark that grades AI on real GitHub issues.
Pick one concrete task so you can judge whether the project fits.
Clone the repo and check the prerequisites listed in the README.
Install what it needs and configure credentials or model access.
Run a minimal example end to end before wiring it into your stack.
Built from genuine GitHub issues, not toy problems.
Widely cited for coding-agent evaluation.
Resolves are verified with real test suites.
Accompanied by published methodology.
Harness and datasets in Python.
Permissive licensing.
Install the SWE-bench package.
pip install swebenchPull the task instances you want to evaluate on.
# see docs for dataset optionsEvaluate a model per the documentation.
swebench --help
# see docs for eval commandsIt is maintained by the SWE-bench organization at SWE-bench/SWE-bench, with 1,004 forks.
Yes. It is released under the MIT license, which permits commercial use, modification, and redistribution. Read the LICENSE file for the exact terms.
Running evals needs model access and compute; the benchmark itself is free.
It targets real-world software engineering; compare with AgentBench or GPQA for different agent capabilities.
Read the paper and site docs, install the harness, and run a small evaluation.
Give it a GitHub issue and it tries to fix the code for you - the original SWE-bench agent
A benchmark for tool-agent-user interaction in real-world domains.
A benchmark testing whether agents can solve long-horizon coding tasks.
GitHub's official MCP server — let AI tools work with repos, issues and pull requests
Self-hosted AI vulnerability research: agents find, validate, deduplicate, and rank real security issues.
Benchmark computer-use agents on long, real-world desktop tasks.
LibreAIカタログに掲載されたAIエージェントプロジェクトの中で最もスターの多いもの — SWE-benchが合わなかった場合に検討したい、現実的な候補リストです。
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
カタログの中でSWE-benchとスター数が近いAIエージェントプロジェクト — 前後の棚もそのまま見て回れます。
A desktop-grade AI agent from NetEase Youdao that does real work — analysis, slides, docs, video and research.
An open-source auth gateway connecting 1,500+ SaaS providers to AI agents via SDK, CLI, MCP and HTTP.
Open-source AI coworkers, each with a computer of their own.
A framework for evaluating and improving AI agents.
An Agent OS with interview-gated, budgeted evolution — one MCP server across 14 agent runtimes.
A zero-code platform for auto-generating production-grade AI agents with unified tools, skills and memory.