Usage 0.29.1

Released: 2026-07-25

Release notes

FEATURE

Give every Council seat its own expert hat

In one Council (a feature where several AIs discuss an issue from separate seats), each seat can now take a different specialty. Need one person to review a contract, one to spot front-end trouble, and another to audit security? A role picker beside the model picker lists specialists from your installed talent packs. A persona (an expert hat placed on an AI before it speaks) is injected as a prompt prefix, so the same CLI (a tool launched from a command line) can wear two different hats in two seats. Roles are grouped by talent pack in a collapsible picker, keeping even 50 choices manageable. Their text is fetched from the talent-market CLI at runtime instead of bundled into usage: like casting a meeting, each chair gets the right specialist.

FEATURE

Choose a debate style, then stop once everyone agrees

When a group of AIs keeps debating, you can now decide whether the meeting should brainstorm together or hunt for flaws. From round two onward, Council lets each participant take one of five debate styles: constructive consensus-seeking, adversarial fault-finding, collaborative gap-filling, Socratic questioning of assumptions, or devil's advocate. Whatever style you choose, the stance label opening each reply stays visible and is counted above the summary. Enable end on consensus (a switch that stops once everyone agrees), and the discussion ends as soon as every participant marks agreement in a round instead of using all budgeted rounds. If anyone fails to finish that round, it does not count. Like a chair setting meeting rules, then checking every raised hand before calling the room settled.

FEATURE

Moderators no longer see names, keeping summaries impartial

When a moderator is asked to draw a conclusion from several AIs, it now sees the opinions before it sees their authors. The transcript (a record preserving every spoken line) handed to the moderator is anonymized, so it cannot favor answers written by its own CLI (a tool launched from a command line). Names return before the summary reaches your screen or is exported as Markdown (plain text that can carry formatting), so you still read “Claude argued...” rather than “Participant A.” Only author labels are hidden; the words themselves remain untouched, even when the discussion is about Claude or Codex. It is like a blind wine tasting: judges taste what is in the glass before learning its label.

Original changelog

Original changelogGive every Council seat its own expert hat
**Council participants can wear a persona from the AI Talent Market**: every seat now has a role picker next to its model picker, listing the specialists from your installed talent packs (contract review, front-end, security audit, and the rest). The role's expertise is injected as a prompt prefix, so the same CLI can sit in two seats wearing two different hats. Roles are grouped by talent pack in a collapsible picker — collapsed it shows only the pack names, so a 50-role list no longer scrolls off the screen. Role text is fetched from the talent-market CLI at runtime and is never bundled into usage itself.
Original changelogChoose a debate style, then stop once everyone agrees
**Five debate styles for a council**: pick the tone every participant takes from the second round on — constructive consensus-seeking, adversarial fault-finding, collaborative gap-filling, Socratic questioning of assumptions, or devil's advocate. The stance label each participant must open with is preserved under every style.
**Council consensus is counted, and can end the discussion early**: the stance labels participants open their rebuttals with are now tallied and shown above the summary. Turn on *end on consensus* and a discussion stops as soon as every participant in a round agrees, instead of burning through the rounds you budgeted for. A round where anyone failed to finish never counts as consensus.
Original changelogModerators no longer see names, keeping summaries impartial
**The moderator no longer sees who said what**: the transcript handed to the moderator is anonymized, so it cannot favor the answers its own CLI wrote. Participant names are restored before the summary reaches your screen and the exported Markdown, so you still read "Claude argued…" rather than "Participant A argued…". Participants' own words are passed through untouched — a discussion *about* Claude or Codex stays readable.