jj-workflow

Automating jj version control tasks requires a non-interactive workflow that enforces mandatory -m commit messages for every operation. This approach replaces interactive staging with a mental model where @ is the current change and @- is the parent, streamlining automated commit descriptions and pushes.

The no-staging mental model in jj treats @ as your current change and @- as its parent, eliminating the need for traditional staging areas. This model enforces commit-based workflows by requiring descriptive -m messages for operations, ensuring every change is safely recorded without manual intervention.

Recovering changes in jj after a bad squash or describe operation uses the oplog-based recovery mechanism. By accessing the operation log, you can restore previous states of your repository, ensuring safe commits are preserved even when non-interactive commands modify or squash changes unexpectedly.

JJ workflows support colocated repository setups identified by a .jj directory or system reminders indicating vcs=jj-colocated. This skill applies non-interactive commands like new, describe, squash, and push across branches and remotes seamlessly within both standard and colocated environments.

JJ requires a descriptive message for every commit operation to enforce a strict non-interactive workflow. By mandating -m flags for commands like new, describe, and squash, the system ensures clear commit descriptions and maintains a safe, automated version-control history without user prompts.

A non-interactive jj workflow limits manual intervention during commit creation, squashing, and describing. It requires mandatory -m messages for all operations and relies on oplog for recovery, meaning users cannot selectively stage files interactively and must adapt to the @ and @- mental model for all version control tasks.