Agent skills
Install the Anzen agent skill with the open skills CLI so coding agents get consistent guidance on server actions, route handlers, and Server Components.
Agent skills are reusable instruction sets for coding agents (Cursor, Claude Code, Codex, and many others). They ship as SKILL.md files with YAML frontmatter and are installed with the npx skills CLI. Browse and discover skills at skills.sh.
This repository includes an Anzen skill so agents apply the same patterns as these docs: Standard Schema validation, authorize behavior per surface, and when factories are worth using.
Install the skill
npx skills add SugarDarius/anzenIf the install source exposes several skills, select this one with --skill anzen.
What’s in the repo
| Path | Role |
|---|---|
skills/anzen/SKILL.md | When to use Anzen, steps, imports, validation rules, fair use |
skills/anzen/references/ | Short topic files (validation, authorization, each factory) with links back to this documentation |
The skill’s description field is written so agents can decide when to load it (Next.js App Router, @sugardarius/anzen, safe actions/routes/RSC, and so on).
References inside the skill
SKILL.md holds the overview: when to use Anzen, a short workflow, imports, validation and authorization summaries, fair use, and a catalog of reference files. The references/ folder splits deeper material by topic so agents (and humans) can open only what matches the task instead of one long page.
Each reference file is plain Markdown: compact tables for options and behavior, plus links to the matching API page in this site’s source (www/content/docs/*.mdx) on GitHub for full prose, signatures, and fair-use notes.
| File | What it covers |
|---|---|
validation.md | Standard Schema, synchronous validate, dictionary vs single-object schemas, body vs formData on route handlers |
authorization.md | How authorize works for route handlers (including Response), server actions, and RSCs; what each callback receives |
server-actions.md | createSafeServerAction options, handler context, structured results and error codes |
route-handlers.md | createSafeRouteHandler options, validation callbacks, (ctx, req) signature |
server-components.md | Page vs layout factories, shared base options, experimental_slots |
Cross-links between reference files (for example route handlers pointing at validation) stay one level deep under references/, which keeps navigation predictable for tools that load skills incrementally.
Last updated on