What problem does it solve? AI agents working in repositories managed by Jujutsu (jj) often hang on interactive pagers and editor prompts, misuse git commands, or produce messy commit histories. This Skill provides the exact command patterns and safety rules for operating jj correctly in automated environments. ## Core Features & Use Cases - Agent-Safe Command Patterns: Enforces --no-pager and inline -m message flags so commands never block on interactive prompts. - Mutable Commit Workflows: Covers describing, squashing, absorbing, rebasing, splitting, and abandoning commits to keep history atomic and clean. - Bookmarks, Workspaces, and Git Interop: Explains bookmark management, multi-workspace handling, and safe git fetch/push integration in colocated repos. - Use Case: An agent asked to implement a feature uses jj desc -m to set the commit message first, edits files, verifies with jj st, and refines the commit with jj squash or jj absorb before finishing. ## Quick Start Use the jujutsu skill to create a new commit describing my change, then show me the current repository status and recent log.