What problem does it solve? Running marketing campaigns with AI often produces generic, inconsistent output because agents lack brand context, structured workflows, and memory of past results. This Skill sets up a coordinated team of four specialized agents (Orchestrator, Researcher, Strategist, Copywriter) that pass work through JSON schemas and retain campaign history across sessions. ## Core Features & Use Cases - Multi-Agent Pipeline: An Orchestrator routes requests to a Researcher (audience profiles), Strategist (creative briefs), and Copywriter (finished copy), communicating via structured JSON contracts. - Persistent Memory: Campaign history, brand voice examples, and task queues are stored in markdown memory files so the system improves and resumes work across sessions. - Built-in Frameworks: Applies 6 hook formulas and 7 growth playbooks to every deliverable, with automatic brand voice enforcement per client. - Use Case: An agency managing multiple clients asks for a campaign targeting marketing directors at B2B SaaS companies; the system produces an audience profile, a creative brief, and five LinkedIn posts plus a landing page hero, all logged to campaign history. ## Quick Start Clone the marketing-os-starter repository, fill in CLAUDE.md and your agency brand files, then ask Claude Code to build a campaign for your first client with a target audience, budget, and timeline.
