CURRENT / Sources & referenceSource status

Guides / Agent skills

Give your agent a repeatable method.

Reusable instructions for scoping, building, reviewing, and coordinating work. Each skill includes an explanation and files to take with you.

01

Product thinking · 13 min read

Turn an idea into a buildable brief ↗

Work through a reading-list feature from a vague request to a finished brief, precise behavior, and the first executable tests.

Take away: A completed product brief, a URL policy with tests, and a reusable skill for the next ambiguous request.

7 chapters3 example filesSKILL.md included
02

AI engineering · 12 min read

Build an AI feature with explicit failure behavior ↗

Build a support-ticket triage core that can say “I could not complete this,” survive duplicate requests, and stop before exhausting its allowance.

Take away: A runnable triage core, seven tests of failure and contract behavior, and a plan for taking the same boundaries into a database-backed service.

7 chapters2 example filesSKILL.md included
03

Design & UX · 13 min read

Review an interface through real tasks ↗

Follow one reader through a save, a duplicate, and a failed request. Turn the findings into a working form, recovery behavior, and repeatable tests.

Take away: An interactive reading-list lab, its complete HTML source, a tested state transition, and a worked review worksheet.

7 chapters4 example filesSKILL.md included
04

Agent workflows · 12 min read

Coordinate agents without overlapping ownership ↗

Split a reading-list feature between API and interface work, make their agreement executable, and integrate the result without relying on reassuring handoffs.

Take away: A concrete ownership plan, a shared request/response contract, four executable contract checks, and an integration procedure.

7 chapters4 example filesSKILL.md included