Runs Ian's hardest tasks with a lead and a crew of up to 8 agents, after the Fable 5.2 multi-agent post: each writer on its own branch, nothing counts until a check passes, a read-only verifier tries to refute every result, and a guard refuses any deploy, push, send, secret read or bulk delete during the run and halts the whole crew until Ian reviews it.
Turns a rough prompt into a clear one: up to three quick questions, a rewrite, and a side-by-side review before anything runs.
Runs each skill and subagent on the right size of model, picked by Jev through OpenRouter and never above the session model. Shows each running subagent in a band above the chat box, with its model by colour and a progress bar, and on an Agents pane with how Jev chose the model. Gives each session conductor rules for handing work to subagents, and has a Fable agent review any turn that changed...
Counts the tasks Claude finishes for you and estimates the hours saved and what that time is worth, by day, week, month and year.