Turn GitHub issues into reviewed code changes with SWE-agent
Run SWE-agent against a repository issue so a language model can inspect files, execute commands and tests, and produce a patch for human review.
npx skills add agentskillexchange/skills --skill turn-github-issues-into-reviewed-code-changes-with-swe-agent
This skill helps maintainers run an autonomous software-engineering agent on a concrete GitHub issue or repository task. SWE-agent gives the model tools to inspect the codebase, edit files, run commands and tests, and iteratively work toward a proposed fix.
What this skill actually does
Invoke it when the desired output is a reviewable patch for a scoped issue, benchmark task, or repository repair workflow rather than an interactive IDE pairing session. The boundary is issue-to-patch execution with explicit maintainer review before merge; it is not a generic benchmark entry, research paper card, or unrestricted autonomous coding product.
Inputs and prerequisites: SWE-agent; Python environment; Git repository or GitHub issue; model API credentials; test/runtime dependencies for the target project.
Setup notes: Follow the SWE-agent documentation to install the CLI, configure a model, point it at a repository issue or task, run the agent, then review and test the generated patch before merging.
Source and verification boundary: use https://swe-agent.com/latest/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Custom Agents workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.