v0.48.0

Improvements

• Review agents run in a read-only sandbox to better protect your checked-out worktree.
• Review jobs avoid .git/index.lock contention, improving reliability when Git is active.

Bug Fixes

• Development builds no longer show update notifications.

v0.47.2

Bug Fixes

• roborev fix correctly honors --all-branches and --branch when run from Git worktrees.

v0.47.1

• Improve reliability by isolating Git-related test execution from global Git and GPG configuration.
• Fix a session-stream test agent leak in the global registry to improve overall stability.

v0.47.0

New Features

• Add roborev summary to show aggregate review statistics.
• Add TUI control socket support, runtime metadata, and a --no-quit flag for longer-lived sessions and external control.
• Track agent token usage and add token backfill support for existing data.
• Add an opencode_cmd config override to use a custom executable path.

Improvements

• Exclude common lockfiles from reviews by default and add configurable exclude patterns.
• Add Codex maximum reasoning to support xhigh reasoning mode.
• Show a session ID column in the TUI queue view.
• Let TUI column checkboxes respond to mouse clicks.
• Word-wrap long comment text in the TUI review pane.
• Update the TUI elapsed timer every second.
• Improve skill naming compatibility by switching to hyphenated skill names.

Bug Fixes

• Fix ambiguous branch detection so GetCurrentBranch no longer returns a heads/-prefixed branch name.
• Update Gemini defaults and fall back cleanly when the configured model is unavailable.

v0.46.1

• Ensure roborev fix only discovers reviews for the current Git worktree.
• Restore post-commit hook execution in linked Git worktrees.

v0.46.0

• Honor *_cmd configuration overrides when checking agent availability, so custom agent commands are detected correctly.

• Show the stored branch name in the review screen instead of the active worktree branch.

• Use the main repository root for the post-commit hook when running inside git worktrees.

v0.45.0

• Add --min-severity to roborev fix and roborev refine to target only findings at or above a chosen severity.

• Include review comments when copying a review, making shared output more complete.

• Reuse compatible review sessions on the current branch more safely, preserving context without crossing branch boundaries.

• Add clearer documentation to TOML config files that roborev writes automatically.

• Fix roborev compact in git worktrees so it uses the correct branch.

v0.44.3

Improvements

• Hide daemon status details in the TUI review view to keep review output focused.
• Improve job logging and make timeout cleanup more reliable for long-running daemon work.

Bug Fixes

• Remove tool-call narration from review output so agent reviews stay concise and readable.

v0.44.2

• Clarify review prompts so agents do not run builds or tests during roborev review.

v0.44.1

• review, analyze, fix, and refine now fall back to the selected agent’s own default model when no workflow-specific model is configured.

• Queued and CI-triggered review jobs use the correct agent/model fallback behavior for the configured agent.