Skills for your AI
Methods you load into the AI assistant you already use. When you ask for something the method covers, it follows it step by step instead of improvising.
-
01
€9.99
Product research and validation
Tells you when your idea isn't the best.
A step-by-step wizard: goal, market data, opportunity ranking, benchmark, voice of the user, findings, insights. Every data point needs a source and a reliability score, or it gets labeled as an assumption.
-
02
€9.99
Design system and architecture
A system that doesn't drift, or break silently.
Two parts. A token-based design system (type, WCAG contrast, components, navigation, glossary) and a data architecture that survives production (CloudKit schema, versioning, fast QA).
How a skill works
A text file, nothing magic
A skill is a SKILL.md file: instructions in Markdown with a name and a short description at the top. The format started with Claude and is now an open standard, Agent Skills, read by many AI tools. The assistant reads the description, recognizes when it applies, and follows the method.
- Tools with skills support (Claude, OpenAI Codex, Cursor, GitHub Copilot, Gemini CLI and others): put the pack folder in the tool's skills directory, or upload it as a skill. For Claude Code that's
~/.claude/skills/. - Chat assistants (ChatGPT, Gemini): paste
SKILL.mdas instructions in a custom GPT, a Project or a Gem, and attach the worked example. - No AI at all: it still reads as a step-by-step checklist.
Support varies: in some tools the method starts on its own, in others you ask for it by name. And a skill doesn't make any AI infallible. It gives you a repeatable method and stops you from skipping the boring steps, which are usually the ones that matter.
Each skill comes with its worked example and bonuses, all in the same download.