# y8y skills

Four free skills for AI agents (Claude Code, Claude.ai and any tool that
reads `SKILL.md`), distilled from systems running in production at
[y8y.ai](https://y8y.ai): a trilingual robotics publication operated with AI
agents, with costs, governance and failure logs published in the open.

No fabricated numbers, no hype. Every skill states its method and links to
the live system it was distilled from.

| Skill | What it does |
|---|---|
| [honest-case-study](skills/honest-case-study/SKILL.md) | Writes case studies readers and AI engines trust: six beats including honest costs, measured results and what failed |
| [ai-use-case-matrix](skills/ai-use-case-matrix/SKILL.md) | Turns an AI wish list into two measured pilots: task inventory, 3-criteria scoring rubric, pilot charter with a scale-or-kill date |
| [research-integrity-contract](skills/research-integrity-contract/SKILL.md) | Makes agent research citable: verbatim-or-empty extraction, URL per row, frozen statistics, independent recount |
| [honest-ai-cost-account](skills/honest-ai-cost-account/SKILL.md) | Computes the real cost of an AI workflow: tokens + platform + the supervision line everyone omits |

## Install

**Claude Code**: copy any skill folder into `.claude/skills/` in your project
(or `~/.claude/skills/` for all projects). Invoke with `/honest-case-study`,
or let Claude trigger it from context.

**Claude.ai**: upload the `SKILL.md` as a skill in your workspace settings.

**Anything else**: `SKILL.md` is plain markdown with YAML frontmatter; any
agent that reads the emerging skill format can use these as-is.

## Why these four

They cover the two failure modes of most AI work: claiming results nobody can
verify, and hiding what things really cost. The formats were built against an
audit of about 50 top-ranking pages in three languages where zero disclosed
costs, supervision hours, measurement method, governance or failures.

## Who is behind this

[Sebastián Ocampo](https://y8y.ai/es/autores/seb-ocampo/), founder and editor
of y8y and PECB Certified AI Manager (CAIM). He helps companies grow with
well-governed AI systems; the publication is his workshop in the open. The
long-form versions of these skills, with worked examples, failure logs and
FAQ, live in the [AI systems lab](https://y8y.ai/es/lab/sistemas-de-ia/).

Contact: hola@y8y.ai · [LinkedIn](https://www.linkedin.com/in/seb-ocampo/)

## License

MIT. Use them, adapt them, ship them. Attribution appreciated, not required.