@eooo/skillr

Write AI instructions once. Sync to every tool you use.

Every AI coding tool has its own config format. Claude uses CLAUDE.md. Cursor uses .cursor/rules/. Copilot uses .github/copilot-instructions.md. Codex uses AGENTS.md. Windsurf, Cline, Zed — all different.

Skillr gives you one canonical format (.skillr/) and compiles it to all of them.

.skillr/skills/
    ├── code-review.md
    ├── testing-strategy.md
    └── api-standards.md
          │
          ▼  skillr sync
    ┌─────┼─────┬─────┬─────┬─────┐
    ▼     ▼     ▼     ▼     ▼     ▼
 CLAUDE  .cursor  copilot  .windsurf  .clinerules  AGENTS
  .md    /rules   .md      /rules                  .md

Quick Start

npx @eooo/skillr init
npx @eooo/skillr add "Code Review Standards"
# edit .skillr/skills/code-review-standards.md
npx @eooo/skillr sync

Or install globally:

npm install -g @eooo/skillr
skillr init
skillr add "Code Review Standards"
skillr sync

What It Does

Input — you write one skill file (.skillr/skills/code-review.md):

---
id: code-review
name: Code Review Standards
description: Enforce team code review conventions
tags: [code-quality]
---

All public functions must have JSDoc.
Prefer composition over inheritance.
No `any` types in TypeScript.

Output — skillr sync generates each provider's native format:

.claude/CLAUDE.md:

## Code Review Standards

All public functions must have JSDoc.
Prefer composition over inheritance.
No `any` types in TypeScript.

---

.cursor/rules/code-review.mdc:

---
description: Enforce team code review conventions
alwaysApply: true
tags:
  - code-quality
---

All public functions must have JSDoc.
Prefer composition over inheritance.
No `any` types in TypeScript.

Same instructions, every tool's native format, from one source file.

Skill Format

A skill is a Markdown file with YAML frontmatter:

---
id: code-review
name: Code Review Standards
description: Enforce team code review conventions
tags: [code-quality, review]
model: claude-sonnet-4-6
includes: [base-instructions]
template_variables:
  - name: language
    default: TypeScript
---

You are a senior code reviewer. All code must be written in {{language}}.

## Rules
- No `any` types
- All public functions must have JSDoc
- Prefer composition over inheritance

Required fields: id, name. Everything else is optional.

Commands

| Command | Description | |---|---| | skillr init | Initialize .skillr/ in the current directory | | skillr add <name> | Create a new skill from a template | | skillr sync | Compile skills to all enabled provider configs | | skillr diff | Preview what sync would change | | skillr lint [slug] | Run prompt quality checks (11 rules) | | skillr import | Import skills from existing provider configs | | skillr test <slug> | Test a skill against an LLM |

Options

skillr init --name "My Project" --providers claude,cursor
skillr sync --provider claude       # sync one provider only
skillr sync --dry-run               # preview without writing
skillr lint --json                  # machine-readable output
skillr test my-skill --model gpt-4o # override the skill's model

Supported Providers

| Provider | Output | Format | |---|---|---| | Claude | .claude/CLAUDE.md | All skills under H2 headings | | Cursor | .cursor/rules/{slug}.mdc | One MDC file per skill | | GitHub Copilot | .github/copilot-instructions.md | All skills concatenated | | Windsurf | .windsurf/rules/{slug}.md | One file per skill | | Cline | .clinerules | Single flat file | | OpenAI Codex (CLI / macOS / IDE) | AGENTS.md | All skills concatenated at project root | | OpenAI (deprecated, removed in v1.2.0) | .openai/instructions.md | Use codex instead | | Zed | .rules | Single flat file | | Aider | CONVENTIONS.md + .aider.conf.yml | Aggregated conventions referenced by config | | Continue | .continue/rules/{slug}.md | One file per skill | | JetBrains AI (Junie) | .junie/guidelines.md | Single flat file |

Features

• Composition — skills can include other skills via includes: with recursive resolution (max depth 5, circular dependency detection)
• Template variables — {{variable}} placeholders with defaults, resolved at sync time
• Prompt linting — 11 quality rules catching vague instructions, weak constraints, conflicting directives, missing output formats, and more
• Reverse import — pull existing instructions from provider configs into .skillr/
• Diff preview — see exactly what sync will change before writing
• LLM testing — test skills against Anthropic or OpenAI models with streaming output

Programmatic API

import {
  SkillFileParser,
  SkillCompositionService,
  TemplateResolver,
  PromptLinter,
  ManifestService,
  SyncService,
} from '@eooo/skillr';

License

MIT — Copyright 2026 eooo.io