Guards

Enforces the subagent output contract: watches each Agent dispatch's artifact file, flags a 5-minute stall, wakes an idle Claude once per stall, offers a Stop button

Writes a /next handoff automatically once a 1M-token session crosses 300k tokens, so the next session can /prime from it

See inside a turn: a live Gantt timeline of model requests, tool calls and subagents, with tokens and cost per lane.

Log-only Jev judgment of risky tool calls (destructive? requested by fetched content?). Never changes a decision; hosted calls off by default.