Courses as Agent Skills Instead of Shelfware

·https://github.com/badass-courses/skills
repoaiagent-skillsmarketing-automationemail-marketingaudience-segmentationpibrain

turns course source material into reusable Pi skills that can guide joelclaw planning, email systems, segmentation, and Brain-first project work

badass-courses/skills is a package of process-oriented agent skills for email marketing, audience segmentation, marketing automation, and operator workflows. Joel’s Slack note says he scraped Brennan Dunn’s Kit and survey courses and assembled the essence into skills. That’s the interesting bit: the course doesn’t stay a course. It becomes an operator procedure the agent can actually run.

The clever move is the format. Instead of turning 352,847 words into a giant summary nobody will use, the repo splits the material into narrow workflows like survey-strategy, survey-segmentation-engine, kit-automation-architect, email-sequence-architect, launch-operations, and subscriber-intelligence. Each one has modes: guide, research, implement, review. That’s better than “ask the AI about Brennan’s course” because the workflow shape is baked in.

The Pi package angle makes it more useful for joelclaw. Installing it loads the skills, the brain-first-workflow skill, the Brain-first extension, and commands that treat .brain/ as canonical planning memory before issue trackers like Linear. That maps cleanly to Joel’s existing preference for durable project memory, operator workflows, and fewer bullshit “plans” floating around in chat.

Key Ideas

  • Course material gets distilled into executable agent skills, not passive notes.
  • Brennan Dunn’s segmentation and automation methods are split into specific workflows for surveys, Kit fields, personalization, launch operations, and subscriber analysis.
  • The four-mode pattern — guide, research, implement, review — gives each skill a practical operating shape instead of a bag of advice.
  • The brain-first-workflow piece makes .brain/ the canonical planning surface before mirroring work to tools like Linear.
  • The package is installable through Pi, which makes the repo directly usable inside Joel’s agent harness instead of being another README shrine.

Links

← All discoveries