Working in a Terminal Based Coding Agent
Use a command line coding agent to explore, edit, test and commit code safely in real projects
A taste of a lesson
The agent told me 'all tests pass', but CI failed right after. How can that happen?
A few common causes. The agent may have run only some tests, such as one file, and summarised as if it ran everything. It may have run them before its last edit. Or your local environment differs from CI: versions, environment variables, a database. Fix the habit: ask it to run the full test command from your project instruction file after the final change, and read the output yourself instead of the summary. What command does your CI use to run tests, and is it written in your instruction file?
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- Run a session workflow: branch, orient, plan, implement, review, commit
- Choose permission levels and keep destructive commands behind approval
- Manage context in long sessions with notes files and fresh sessions
- Use tests, linters and type checks as feedback for the agent
- Review agent diffs thoroughly before committing
Lesson plan
- 1 What a terminal agent does Understand the read, edit, run and iterate loop and its risks. Start
- 2 Safety first: branches and permissions Set up version control and permission levels before the first task. Start
- 3 Orient and plan Have the agent understand the code and propose a plan before editing. Start
- 4 Implement with feedback loops Let tests, linters and type checks guide the agent's iterations. Start
- 5 Context in long sessions Keep long sessions focused and accurate. Start
- 6 Reviewing and committing Review every change and commit in small, clear steps. Start
Try asking
About this tutor
For developers comfortable with a terminal who want to work with a coding agent that reads files, edits code and runs commands from the command line. These agents can carry out multi step tasks across a codebase, but they need clear tasks, sensible permissions and a reviewer who reads the diff. You will learn a reliable session workflow: orienting the agent, planning before editing, letting it run tests, reviewing changes, managing context in long sessions, setting permission levels for commands, and using version control as your safety net. Tool neutral, with patterns that apply across agents.
Reviews
4.7
3 ratingsSample
- Ben L.Sample
Tool neutral but concrete. I switched agents halfway through the course and everything still applied.
- Hye-jin K.Sample
Great on permissions and context management. The notes file trick fixed my problem with long sessions forgetting constraints.
- Arturo V.Sample
Branch, orient, plan, implement, review, commit. Writing it on a card sounds silly but my sessions are so much calmer now.
About the teacher
I teach developers and beginners the daily habits that make AI coding assistants useful instead of risky
9 tutors 411 lessons taught Sample
I am a software developer who teaches the working habits around AI coding assistants: how to describe a task, how to pair with an assistant without losing the thread, how to use a terminal based coding agent, and how to write project instruction files that keep it on track. I teach from the keyboard. We open a small project, try...
See Jude's profile and tutorsMore like this
Other tutors on the same or nearby topics.