You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Queen: actor runtime MVP, the reviewer as actors behind a flag, and the benchmark #1697
Owner rule, 2026-10-08 (epic gHashTag/t27#7851): the Queen moves to the actor model. Replacement follows a fixed order:
An MVP.
Its tests.
A benchmark against the current loop on the same input.
The numbers posted.
Only then the swap.
The reviewer is the first domain.
Acceptance criteria
queen-actors.ts runs actors whose every decision comes from the generated card specs/queen/actors.wasm. Those decisions are: delivery, lane order, restart, backoff, intensity and the turn bound.
queen-review-actors.ts runs the reviewer as a supervision tree. It starts only with TRIOS_QUEEN_REVIEWER=actors; the default stays the loop.
tests/api/queen-actors.test.ts passes.
tests/api/queen-actors-bench.test.ts runs the loop and the actors on the same seeded input and prints the table.
Activity
fix(queen): a killed review keeps its row and its key lane until it r…
feat(queen): actor runtime MVP, the reviewer as actors behind a flag,…
feat(queen): the actor runtime MVP, the reviewer as actors behind a f…
fix(queen): a killed review keeps its row and its key lane until it r…
gHashTag commentedon Oct 10, 2026
Done: merged into
queenby #1698.Closesfires only on the default branchmain, so this is closed by hand.