SWE-agent
An open-source agent that takes a GitHub issue and tries to fix it using a language model of your choice.
These buttons open the developer's own site, repository or store listing in a new tab. wares.gg does not host downloads.
About SWE-agent
SWE-agent gives a language model a set of tools for navigating and editing a repository, then points it at a GitHub issue and lets it attempt a fix. It works with the model of your choice and was presented at NeurIPS 2024.
Beyond issue fixing, it can be used for competitive coding challenges and security research. It is run from the command line and suits researchers and developers experimenting with autonomous software engineering agents.
Strengths
- Works with the language model of your choice
- Open source with an active research community
- Designed to resolve real GitHub issues end to end
Limitations
- Research-oriented setup rather than a polished product
- Needs a capable model, which may carry API costs
- Security uses are intended for authorized testing only
Details
- Pricing
- FreeFree and open source; model API usage may cost money.
- License
- Open source, license not stated
- Developer
- The SWE-agent contributors
- Platforms
- Command line
- How it runs
- Downloadable app
- Best suited for
- Researchers and developers experimenting with autonomous issue-fixing agents
- Last verified
- Added
Alternatives to SWE-agent
Compare allSoftware that can replace SWE-agent for an important use case, and what changes if you switch.
Devin
An autonomous AI coding agent that works on engineering tasks and delivers ready-to-merge pull requests.
Devin is a paid, closed hosted agent that takes tasks end to end to ready-to-merge pull requests, instead of an open-source research tool using your own model.
Jules
Google's asynchronous coding agent that works on GitHub repositories in a cloud VM and opens pull requests.
Jules runs issue-driven tasks in a Google cloud VM and opens GitHub pull requests, replacing your own model choice with a freemium hosted Google service.
Factory
AI coding agents called Droids that plan, build, review and ship code from desktop, terminal and web.
Factory offers paid, enterprise-oriented Droid agents across desktop, web and CLI with model choice, rather than a free open-source research setup.
Open Interpreter
A command-line coding agent built to work with open language models such as Kimi and GLM.
Open Interpreter is a free terminal coding agent built around open-weight models, more of a general agent than one focused on resolving GitHub issues.
Aider
AI pair programming in a terminal, working directly on your git repository and committing as it goes.
Aider is a free Apache-2.0 terminal tool that commits each change to git and runs tests, but it is interactive pair programming rather than autonomous issue fixing.
Similar software
Related functionality, not necessarily a direct replacement.
Claude Squad
A terminal app for running several AI coding agents at once, each in its own workspace.
Vibe Kanban
A Kanban board for planning, running and reviewing work done by AI coding agents such as Claude Code and Codex.
CodeRabbit
An AI code review service that comments on pull requests with line-by-line suggestions and chat.
gptme
An open-source AI agent for the terminal that writes code, runs shell commands and browses the web.
Claude Code
Agentic coding tool that reads a codebase, edits files and runs commands.
