SWE-agent
Agent FrameworksSWE-agent/SWE-agent
20.7K2.3KPythonMIT
An AI agent that reads a GitHub issue and then explores the code, edits files, runs tests, and submits a fix on its own.
💻 Code Assistants🏗️ Agent Building⚙️ Automation Workflows🔒 Cybersecurity
Let your AI agent install it
project-specificCopy this into Claude Code, Codex, Cursor, or any assistant that can run commands. It carries this project's own install command and verification step.
Install and configure SWE-agent on my machine. Repository: https://github.com/SWE-agent/SWE-agent What this project is: An AI agent that reads a GitHub issue and then explores the code, edits files, runs tests, and submits a fix on its own. Installation given by the repo's docs (if it's stale, follow the current README): whatever the README prescribes Verify it works with this (also per the repo's docs): whatever minimal example the README gives Prerequisites and caveats: nothing unusual — check the README Requirements: 1. Read the repo's README and install docs first — check whether the commands above still apply and whether dependencies are complete. Don't guess. 1b. **If it turns out not to be installable software** (a dataset, project template, config collection, documentation, design assets, or a phone app), tell me what it actually is and how it should be used, then proceed that way — don't force an install flow. 2. Choose steps for my OS (macOS / Linux / Windows). If unsure, ask me first — no destructive operations. 3. Actually run the verification step above and paste the real output, so we know it runs — not "it should work". 4. Ask before you need an API key, admin rights, or a large download / significant disk usage. 5. Finally tell me in a few lines: how to start it, how to uninstall it, and where the config lives.