Demo Dive Series
You write the same data quality summary every Monday. Or you triage the same alert types every sprint. The steps are clear enough, but the output shifts depending on who runs them, what they remember to check, and how much time they have. A skill turns that workflow into a fixed recipe: one file that tells an AI agent exactly what to pull, how to format it, and what good output looks like.
In this session, we’ll build one skill end-to-end — a weekly data quality report — using Coalesce MCPs. You’ll see how to go from a blank SKILL.md file to a working skill that pulls live data, posts to Slack, and produces the same result whether you run it or hand it to someone on their first on-call rotation. We’ll also run assertions against the output so you can prove it holds up before you share it with your team.
You’ll learn how to:
- Build your first skill from a blank file: Start with a natural-language description, use skill-creator to generate the first draft, then iterate until the output matches what you’d write by hand
- Test that it works consistently: Run the same skill twice and compare results with assertions — does it include a summary, does it deduplicate noisy monitors, does the formatting hold?
- Connect skills into a workflow: See how the weekly report skill hands off to a triage skill that scores issues and drafts Linear tickets, turning a read-only digest into proposed actions
Save Your Seat
Instructors
