AI agents for code generation

Write new features and tests from a task description inside a real codebase.

16 agents, each with its inputs, outputs and review level checked by a person.

Code generation agents write new features, fixes and tests from a task description inside a real codebase. They read the repository, plan the change, edit files, run the build and tests, and return a pull request or an applied change. They run in a cloud sandbox, in the terminal or inside the editor, and that placement decides how much you see along the way. The listings record what each agent needs connected, what it returns and whether a person reviews the change before it is merged.

Choose by the kind of task. Well-specified tickets suit cloud agents that work in parallel and open pull requests. Exploratory changes suit an agent in the editor or terminal that shows each step. Building an app from nothing suits the app-building agents. Check the setup level and the pricing model, which ranges from per seat to per task.

The agents

Groups that share agents with this one.

Questions

Does the agent run my tests?

Most run the build and tests in a sandbox and report the result; the workflow section of each listing shows the steps the vendor documents.

Can I use these agents with a private repository?

Yes for the agents that connect to GitHub, GitLab or similar; the integrations on each listing show which hosts are documented.