v0.63.0

Release 0.63.0

• Add CI quiet-hours throttling for all users.

• Allow selected workloads to bypass quiet-hours throttling.

• Add machine-readable launch receipts to roborev run for automation.

• Tighten roborev-fix skill triggers to prevent unintended activation.

• Infer the branch when enqueueing reviews from detached-HEAD checkouts.

v0.62.1

Release 0.62.1

• Persist completed CI panel metrics and add roborev export ci-metrics for analysis.

• Add a stable JSON contract for version information.

• Wrap long prompt commands by default in the TUI.

• Allow the Codex agent hook to invoke the roborev-fix skill.

• Prevent ACP Gemini backup paths from pairing incompatible models.

v0.62.0

Release 0.62.0

• Add roborev cancel to cancel a single queued or running job.

• Require explicit requests before Codex or Claude Code invokes roborev skills.

• Honor CLAUDE_CONFIG_DIR and CODEX_HOME when installing agent skills.

• Document model-selectable Gemini ACP configuration and common configuration pitfalls.

• Prevent workflow model settings from leaking into ACP agents through paired ACP configuration.

• Use Antigravity print mode only on versions that support it.

• Prevent pre-commit lint checks from automatically modifying files.

v0.61.3

Release 0.61.3

• Repair roborev-managed git hooks automatically when the daemon starts so stale hooks keep working.

v0.61.2

Release 0.61.2

• Show wall-clock elapsed time for panels in the TUI queue view.

• Stop including full prompt text in metadata-only job listings, reducing response size and avoiding unnecessary prompt exposure.

v0.61.1

Release 0.61.1

• Add incremental review export cursors for syncing only new review data.

• Publish documentation Markdown source files for direct access.

• Update CI review comment metadata for better downstream tracking.

• Improve refine documentation with guidance for Agent Hook automation.

• Retry transient CI review attempts when the daemon starts.

• Improve review enqueue metadata collection reliability.

v0.61.0

Release 0.61.0

• Add export support for completed reviews.

• Add a lookahead review type for detecting time-series bias.

• Add Factory Droid hook and skill support.

• Support per-analysis agent configuration.

• Make post-commit hook timeouts configurable, with longer defaults on Windows.

• Improve self-update behavior on slow downloads.

• Document [analyze.<type>] configuration for fieldless review types.

• Repair managed hooks after roborev updates.

• Harden PostgreSQL job sync when review data contains binary or invalid text.

v0.60.0

Release 0.60.0

• Add roborev quickstart for guided setup and automation onboarding.

• Support Codex -c config passthrough overrides.

• Revamp onboarding docs around automation-first workflows and post-commit reviews.

• Prefer managed roborev shims when installing git hooks.

• Reject dirty submodule edits during roborev refine to avoid overwriting local changes.

v0.59.2

Release 0.59.2

• Parse structured JSON review output from Copilot for cleaner findings.

• Include analyze jobs when discovering review results for roborev fix.

• Open Homebrew tap updates through pull requests for safer release publishing.

• Require workflow backup agents to be configured explicitly.

• Stabilize agent invocation for roborev analyze.

• Avoid marking quota-only review panels as failed reviews.

v0.59.1

Release 0.59.1

• Add configurable fix commit metadata for roborev fix and roborev refine.

• Allow panel subagents to be marked flaky so transient panel failures can be tolerated.

• Render Pi job trace output in job logs.

• Show partial panel costs while panel work is still running.

• Expand docs for configuration, GitHub integration, review workflows, and subagent panels.