What problem does it solve? Recurring routines degrade silently: they repeat the same finding for days, flag everything, or go unread while still occupying a plan-limited slot. This Skill scores each routine against nine failure modes drawn from alert-fatigue research, proves what is broken by quoting the routine's own reports, and writes approved replacement prompts. ## Core Features & Use Cases - Evidence-based audit: Scores every routine out of 18 across nine failure modes, diagnosing from report history rather than prompt text, with quoted report lines as receipts. - Approval-gated rewrites: Produces full replacement prompts with memory, escalation, and handoff clauses, requiring explicit approval of the exact text before any update call. - Slot-budget planning: Checks the subscription plan, recommends deleting dead routines before adding new ones, and wires each routine to the Cowork skill that resolves its findings. - Use Case: Your daily report has flagged the same item four days running. Run the audit to get a scored table, a diagnosis quoting the repeated lines, and a rewritten prompt with an escalation rule. ## Quick Start Ask the assistant to audit my routines and fix the ones that keep repeating themselves.
