Agent is partway through a task in your repository when a command it ran fails. Given no further instruction from you, what happens next?

Easycursoragentterminal

Last reviewed 21 Aug 2026

It tries to fix the error itself and carries on, since running commands and repairing failures are part of its loop.Correct answer
It skips the failing step and moves on to the remaining edits, noting the broken command in its closing summary.
It rolls the task back to the most recent checkpoint on its own and reports that the work could not be completed.
It stops and prints the failing command for you to run locally, then waits for you to paste the result back into the chat.

Show answer & explanation

Correct answer: A. It tries to fix the error itself and carries on, since running commands and repairing failures are part of its loop.

Explanation

The principle — Agent reads your codebase, makes changes to files, runs shell commands and automatically fixes any errors it hits. Execution and repair sit inside the loop, not a stage you authorise.

Why the key is correct — A failing command is simply the next piece of information the agent acts on: it reads the output, adjusts and keeps going. That is why the environment matters.

Why the others are wrong — Stopping so you can run the command describes an assistant that can only suggest; Agent has a terminal and reads the output itself. An automatic rollback does not happen: restoring a checkpoint is always your decision. Skipping the failure would defeat the self-checking.

Remember this — Assume commands will run and errors will be fixed unattended, and choose the working environment accordingly.

Sources — Cursor's AI pair programming help documentation.

Sources

Primary sourceHow does AI pair programming work in Cursor? — Cursor · accessed 19 Aug 2026
“It reads your codebase, makes changes to files, runs shell commands, and automatically fixes any errors it hits.”

Practise 10 questions on this topic

Take Cursor AI Features — Timed Test 1 (10 questions) — scored instantly, explanation for every question, no login.

Start the test

Agent is partway through a task in your repository when a command it ran fails. Given no further instruction from you, what happens next?

Easycursoragentterminal

Last reviewed 21 Aug 2026

It tries to fix the error itself and carries on, since running commands and repairing failures are part of its loop.Correct answer
It skips the failing step and moves on to the remaining edits, noting the broken command in its closing summary.
It rolls the task back to the most recent checkpoint on its own and reports that the work could not be completed.
It stops and prints the failing command for you to run locally, then waits for you to paste the result back into the chat.

Show answer & explanation

Correct answer: A. It tries to fix the error itself and carries on, since running commands and repairing failures are part of its loop.

Explanation

Why the key is correct — A failing command is simply the next piece of information the agent acts on: it reads the output, adjusts and keeps going. That is why the environment matters.

Remember this — Assume commands will run and errors will be fixed unattended, and choose the working environment accordingly.

Sources — Cursor's AI pair programming help documentation.

Sources

Primary sourceHow does AI pair programming work in Cursor? — Cursor · accessed 19 Aug 2026
“It reads your codebase, makes changes to files, runs shell commands, and automatically fixes any errors it hits.”

Practise 10 questions on this topic

Take Cursor AI Features — Timed Test 1 (10 questions) — scored instantly, explanation for every question, no login.

Start the test

Agent is partway through a task in your repository when a command it ran fails. Given no further instruction from you, what happens next?

Related questions

Agent is partway through a task in your repository when a command it ran fails. Given no further instruction from you, what happens next?

Related questions