What problem does it solve? One-off marketing tasks get forgotten or done inconsistently. This Skill turns recurring marketing work — weekly SEO scans, ad-fatigue checks, churn watches, dunning recovery — into repeatable loops an AI agent runs on a schedule, each with a defined trigger, self-check, state handling, and stop condition. ## Core Features & Use Cases - Loop Catalog: 43 ready-to-adapt loop specs covering SEO, paid ads, social, activation, retention, revenue, and referrals, each defined with a nine-part anatomy (cadence, action condition, purpose, skills used, body, self-check, state, stop condition, output). - Guardrails & Compliance: A two-tier action model that keeps drafting autonomous while gating spend, sends, and publishing behind human approval, with CAN-SPAM/GDPR/FTC/ToS rules and a kill switch. - State & Idempotency: Patterns for cursors, dedupe keys, cooldowns, and run logs stored under .agents/loops/ so loops never double-act or re-contact the same people. - Use Case: A B2B SaaS team sets up a weekly ranking-drop watch that checks priority keywords, diagnoses regressions, and stages a fix report — skipping quietly when nothing moved. ## Quick Start Ask the agent to set up a weekly marketing loop for the outcome you want to protect, such as "set up a weekly loop that watches our SEO rankings and tells me what to fix."