Review Stats for Every Source, Safer CLI Agent Mode
Review Statistics now counts reviews from the CLI, IDE and MCP by source, the CLI shows the limit that applies to you, and agent mode protects sensitive files.
Review Statistics now counts every Optibot review, wherever it runs. The CLI also gets clearer limits, safer agent mode, and a new release.
New
Review Statistics counts every review, by source
Reviews from the CLI, the MCP server, and the VS Code and Cursor extension now count alongside pull request reviews. The Reviews Over Time chart breaks each bar down by source, and these reviews are credited to whoever ran them in Reviews by Member.
Average reviews per seat
Avg Reviews / Seat replaces the per-member average, showing how much each assigned seat is being used.
Improvements
Clearer review limits in the CLI
optibot status now shows the limit you'll reach first and when it resets. CLI and MCP reviews also count against their own limit rather than the editor extension's. No update needed.
Safer agent mode
When the reviewer asks for more context, the CLI keeps credentials and other sensitive files on your machine.
CLI 0.10.1
--jsonnow works for full reviews, not just agent mode.optibot statusexits with an error when your API key is rejected.optibot org switchexplains why it can't switch whileOPTIBOT_API_KEYis set, instead of appearing to succeed.- Agent-only flags passed without
--agentnow show a warning.
Reliability
- Faster, more accurate replies on pull requests with long conversations.
- Re-reviews no longer fail when a fix includes binary files.
- Cleaner, more consistent wording in summaries, release notes, and CI comments.
- Out-of-credit messages in the CLI and MCP server now include a link to your billing page.