⚡ Imperat AI Skill

Generate correct, version-aware Imperat commands directly from Claude — no more guessing at the API.

A Claude Code skill that bundles the full v3 and v4 Imperat documentation and teaches Claude to:

1. Auto-detect which Imperat version your project uses (by reading your pom.xml or build.gradle).
2. Read only the docs that matter for your task (annotations, platforms, argument types, etc.).
3. Emit a complete command class in Java or Kotlin with the right annotations, registration call, and platform-specific glue.

---

🚀 Quick install

Pick the path that matches your setup. Most people want Option A.

Option A — Drop it in your global skills folder (recommended)

macOS / Linux

git clone https://github.com/MeveraStudios/Imperat-AI-Skill.git \
  ~/.claude/skills/imperat-commands

Windows (PowerShell)

git clone https://github.com/MeveraStudios/Imperat-AI-Skill.git `
  "$env:USERPROFILE\.claude\skills\imperat-commands"

That's it. Restart Claude Code (or just open a new session) and the imperat-commands skill shows up in /skills.

Option B — Project-local skill

If you want the skill scoped to a single project (e.g. shared with your team via the repo):

git clone https://github.com/MeveraStudios/Imperat-AI-Skill.git \
  .claude/skills/imperat-commands

Commit .claude/skills/imperat-commands/ and every teammate using Claude Code in that repo gets the skill automatically.

Option C — Manual download (no git)

1. Click Code → Download ZIP above.
2. Unzip and rename the folder to imperat-commands.
3. Move it into either:
   • ~/.claude/skills/ (global), or
   • <your-project>/.claude/skills/ (project-local).

---

✅ Verify it loaded

In Claude Code, type:

/skills

You should see imperat-commands in the list. If not, double-check the folder path — Claude Code looks for skills under .claude/skills/<name>/SKILL.md (project) or ~/.claude/skills/<name>/SKILL.md (global).

---

🎯 Try it out

Open Claude Code in any project that already declares an Imperat dependency, and ask things like:

You say…	Claude does…
Create a /heal command for Bukkit with a player argument and a permission node.	Reads First-Command/CreateYourFirstCommand.mdx, Arguments Masterclass/Supported-Types.mdx, Extras/@Permission.mdx, Platforms Guide/Bukkit.md. Emits the full class + registration.
Add a 5-second cooldown to my /report command.	Reads Extras/@Cooldown.mdx, edits your existing class.
Make a JDA slash command that mutes a user for N minutes.	Reads Platforms Guide/JDA.md + Extras/@Range.mdx. Emits a Discord-flavoured slash command.
Convert this nested @SubCommand mess into one @PathwayCommand.	Reads First-Command/@PathwayCommand.mdx and rewrites your file in place.

The skill auto-detects whether you're on v3 or v4 by reading your build file. No flag needed.

---

🧠 How it works

imperat-commands/
├── SKILL.md               ← entry point Claude reads first
└── references/
    ├── v3/
    │   ├── INDEX.md       ← table-of-contents + task→file routing
    │   ├── Introduction/
    │   ├── First-Command/
    │   ├── Arguments Masterclass/
    │   ├── Execution-Pipeline/
    │   ├── Advanced/
    │   ├── Extras/        ← @Permission, @Cooldown, @Async, …
    │   ├── Dependency-Injection/
    │   ├── Kotlin/
    │   └── Platforms Guide/  ← Bukkit, BungeeCord, Velocity, Minestom, JDA, Hytale, CLI
    └── v4/
        └── (same layout)

When the skill triggers, Claude:

1. Detects the version — greps pom.xml / build.gradle* for studio.mevera:imperat-core:<version>. Falls back to v4 if nothing's found.
2. Opens INDEX.md for the matched version. The index groups every doc by category and lists short "if the user asks X, read Y" routes.
3. Loads only the docs needed — never all 50 — and writes the answer.
4. Cites the docs it consulted at the end of the reply, so you can verify or dig deeper.

This is progressive disclosure: the skill itself is tiny (one SKILL.md file), and the 100+ bundled docs only enter context when actually needed. Token cost stays low.

---

❓ FAQ

Does this work outside Claude Code (e.g. Claude.ai, the API)?

Yes — Skills work in any Claude harness that supports them (Claude Code, Claude Agent SDK, Claude.ai's skills feature). The install location differs per harness; consult Anthropic's skills docs for non-CC platforms.

How do I update to the latest docs?

cd ~/.claude/skills/imperat-commands
git pull

The bundled docs mirror docs.mevera.studio/docs/Imperat — a git pull here picks up any doc updates we've published.

What if my project uses Imperat v2 or older?

Not supported — the skill only ships v3 and v4 references. v3 is the current default; v4 is in active development.

Can I force a specific version?

Yes — just say so in your prompt: "Create this command targeting Imperat v4". The explicit mention overrides auto-detection.

Where does the skill look for the dependency?

In order: pom.xml, build.gradle, build.gradle.kts. It looks for any artifact matching studio.mevera:imperat-*:<version> and parses the leading major number.

Does the skill modify my files?

Same as Claude Code's default behaviour — Claude proposes edits/files and you approve them. The skill itself doesn't bypass any of that.

---

🛠️ Contributing / updating bundled docs

The docs under references/v<N>/ are mirrored from the MeveraDocs site. To refresh them:

# from MeveraDocs repo root:
cp -r docs/Imperat/v3/* /path/to/Imperat-AI-Skill/references/v3/
cp -r docs/Imperat/v4/* /path/to/Imperat-AI-Skill/references/v4/

Then commit + push. PRs welcome for SKILL.md improvements (clearer routing, better examples, additional task shortcuts in INDEX.md).

---

🔗 Related

• 📘 Imperat docs: https://docs.mevera.studio/docs/Imperat
• 💻 Imperat source: https://github.com/MeveraStudios/imperat
• 💬 Discord: https://discord.mevera.studio
• 📖 Claude Skills: https://docs.claude.com/en/docs/agents-and-tools/agent-skills

---

Built by Mevera Studios · MIT licensed