Skip to content

AI Editor Setup

Your AI coding editor writes better JarvisCore code when it knows the framework. JarvisCore ships that knowledge as a skill: correct API contracts, the profile decision rule, configuration essentials, and the mistakes we see most, maintained in the same repo as the code it describes.

Install the skill

From your project root:

jarviscore init --skill

This writes SKILL.md to the locations editors discover:

Editor Location Loaded
GitHub Copilot .github/skills/jarviscore/ Automatically, when a task matches the skill description
Claude Code .claude/skills/jarviscore/ Automatically, same trigger model

Commit these files. Everyone who opens the project gets an editor that knows JarvisCore.

Cursor and other AGENTS.md tools

Tools that read AGENTS.md (Cursor, Codex, and others) get the same knowledge with one line in your project's AGENTS.md:

When working with jarviscore code, read .github/skills/jarviscore/SKILL.md first.

Any tool: llms.txt

The documentation site serves llms.txt, a curated index of these docs for AI ingestion. Point research agents or custom tooling at it instead of scraping.

Keeping it current

The skill is versioned with the package. After upgrading JarvisCore:

jarviscore init --skill --force