AI instructions in itomProject

This folder contains the tooling-specific entry points. The content lives in AGENTS.md files next to the code.

itomProject/
├─ AGENTS.md                       # single source of truth (tool independent)
├─ CLAUDE.md                       # stub -> AGENTS.md
├─ .github/
│  ├─ copilot-instructions.md      # stub -> AGENTS.md (always loaded by Copilot)
│  ├─ instructions/*.instructions.md   # path scoped rules (applyTo globs)
│  ├─ prompts/*.prompt.md          # reusable, invocable tasks
│  └─ skills/<name>/SKILL.md       # agent skills
├─ itom/AGENTS.md                  # + itom/CLAUDE.md, itom/.github/copilot-instructions.md
├─ plugins/AGENTS.md               # + plugins/CLAUDE.md, plugins/.github/copilot-instructions.md
└─ designerplugins/AGENTS.md       # + .../CLAUDE.md, .../.github/copilot-instructions.md

Why the duplication?

itom, plugins and designerplugins are independent git repositories (submodules). Instruction files are discovered relative to the workspace root that the AI tool opens:

• opening itomProject → the top-level AGENTS.md / .github/** is used, and the *.instructions.md files add module specific rules via their applyTo globs
• opening a submodule alone → only that submodule's own files are found

Therefore the module rules are maintained inside the submodule and the files under .github/instructions/ only reference them. Keep them short so they do not drift.

Conventions

• Tool independent content → AGENTS.md (supported by Copilot, Codex, Cursor, Jules, ...)
• CLAUDE.md and copilot-instructions.md are thin pointers, not copies (no symlinks — they are fragile on Windows/Git)
• SKILL.md uses the same format for Copilot (.github/skills/) and Claude (.claude/skills/); mirror the folder if you use Claude Code locally
• All files: end with a single newline, no trailing whitespace, no BOM (pre-commit)