AI GitHub
Codex CLI for repository work from the terminal
Use OpenAI's terminal coding agent to inspect, edit and test a project.
Key Takeaways
- The CLI works in a chosen workspace and uses tools for file and command operations.
- Use the project directory and the permissions appropriate for that session.
- Ask it to trace one normal entry point before changing code.
- Access settings and the available model affect what a session can do.
How it works
The CLI works in a chosen workspace and uses tools for file and command operations. Sandbox and approval settings control how those operations reach the surrounding system.
What you need
Use the project directory and the permissions appropriate for that session. A reliable validation command gives the agent and reviewer the same target after edits.
What to try first
Ask it to trace one normal entry point before changing code. Review the proposed change on a separate branch and run the smallest test that would fail without it.
Test how a session handles existing uncommitted changes. Review the final Git status as well as the requested patch so you can spot edits that belong to another task.
Where to look for failures
Access settings and the available model affect what a session can do. Treat a reported successful command as one piece of evidence alongside the diff and a reproduction of the requested behavior.
Related Franklin pages
Use the same clean checkout and failing test to compare terminal agents. Look for correct file selection, a minimal diff and evidence from the test run.
- Make model-assisted edits in a Git checkout and review the result as a diff. Aider for terminal-based pair programming
- Work with a coding agent in the terminal using a configured model provider. OpenCode for provider-flexible coding sessions
For a library of development workflows that can complement a coding agent, see Franklin’s Superpowers profile. Superpowers
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗