Case study · AI tooling · Developer experience

AI workflows for a product org

I treated the team's AI setup as a product. I packaged our conventions into Claude Code plugins and skills, so any engineer or PM gets the same rules, setup help and path from ticket to PR in one install.

The problem

As more of Maxio started building with AI coding agents, each person's setup was different. Getting six apps running locally meant asking around, PR conventions depended on who opened the PR, and non-engineers who wanted to build had no clear place to start.

How I approached it

I built the internal plugins and skills as products in their own right, versioned and reviewed, with install steps a PM could follow. I used them myself every day, so each rough edge I hit became the next fix.

What I delivered

AI tooling · Developer experience

Period: Apr – Oct 2026
Role: Builder and maintainer
Stack: Claude Code · MCP · Bash · Python
Merged PRs: 30+

Line item	Detail
Team rules plugin	An org-wide plugin that enforces draft-only PRs, meaningful branch names and complete PR descriptions in every Claude session.
Ticket to PR	A skill that finds the ticket from the branch name, fills in the repository's PR template, and opens a draft PR from one sentence.
Local setup for 6 apps	A plugin that walks engineers and PMs through running six products locally, syncing data between two of them, and diagnosing CI failures.
Functional-test skill	Tests any Maxio feature end to end through the API and a real browser. It deploys the branch to a dev environment, refuses to run against production, and publishes a test report on the PR.
Always-current skills	A skill that sends each request to the latest version of the right skill, pulled live from GitHub, so non-engineers never have to reinstall anything.
Planning pipeline	Wrote deep-dives and milestone plans for four epics in an AI-assisted pipeline that runs from discovery to delivery. Agents research and draft, and people make the decisions in PR review.

ResultSetup that used to mean asking around became a single install. The plugins are now used and extended by Maxio's engineers.

← All case studies · Next: Segment-level custom pricing →