liveBook · Manning

Goal command

As explained in Claude at Work

The /goal command instructs Claude Code to continue working until a specified condition is reached.

The /goal command tells Claude Code to keep working until it judges a specified condition met. It is useful for a long task in an active session where Claude already has relevant project context. section 5.10

Set a goal and a test gate

After discussing an authentication MVP, the prompt was: “go for it /goal get the mvp working and tested with unit and integration tests”. Claude Code’s evaluator checks progress and can prompt another turn if it judges the goal incomplete. Ambiguous goals can consume more turns and tokens; add a limit such as “don’t go beyond 10 turns” and review the results. section 5.10

How Claude at Work covers it

How deep it goes
Describes evaluator-checked repeat turns and an optional turn cap, without detailing evaluator internals.
Examples
Uses an authentication MVP prompt that asks Claude to deliver working software with unit and integration tests.
What it assumes
Assumes an active session with project context established through design and codebase review.
What you can do after
Operate Claude Code by assigning a long task, setting a turn limit, and reviewing reported checks.
How it connects
Connects goal setting to unit testing and integration testing as completion gates.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Claude at Work. The section links point to the parts of the book it draws on.