# Configuring the i-have-adhd Skill for Specific Project Types vs Global Usage

> Master the i-have-adhd skill by choosing project-specific or global configuration. Learn how the CLI prioritizes local skills for efficient workflow management.

- Repository: [Ayoub Ghriss/i-have-adhd](https://github.com/ayghri/i-have-adhd)
- Tags: how-to-guide
- Published: 2026-08-05

---

**The i-have-adhd skill supports both project-scoped and global installation via the `npx skills` CLI, with loading priority given to local `./skills/` directories over user-wide agent directories.**

The **i-have-adhd** repository by `ayghri` provides a productivity-focused coding skill that helps AI assistants structure responses with action-first, numbered steps. Understanding how to scope this skill—whether to a single repository or across all your projects—is essential for effective adoption.

---

## How Agent Skills Load: The Two-Scope Model

AI agents (Claude, Copilot, Cursor, OpenCode, and others) scan specific directories at session startup to discover skill files. The `i-have-adhd` repository follows the **Agent Skills** specification, which defines two distinct scopes:

| Scan Location | Scope | Loaded When |
|-------------|-------|-------------|
| `./skills/` inside current repository | **Project-only** | Working within that specific repository |
| `~/.<agent>/skills/` or `~/.agents/skills/` | **Global** | Any project the agent opens |

The core skill definition lives in [`skills/i-have-adhd/SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md), which contains the front-matter (`name`, `description`, `disable-model-invocation`, etc.) and the behavioral rules the agent follows.

---

## Project-Specific Configuration

Project-scoped installation keeps the skill isolated to one codebase. This is ideal when you need customized rules for a particular team, framework, or workflow.

To install for the current repository only:

```bash
npx skills add ayghri/i-have-adhd

```

This copies the skill into `./skills/i-have-adhd/` within your repository. The agent loads this local version whenever you work in that directory.

According to the repository's [`SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/SKILL.md), the front-matter setting `disable-model-invocation` (found at lines 10-12) controls whether the skill triggers automatically or awaits explicit invocation.

### Overriding Global Settings Per Project

If you have a global install but need project-specific customization, place a modified [`SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/SKILL.md) at:

```

./skills/i-have-adhd/SKILL.md

```

The agent prioritizes local skills, so your customized version overrides the global one for that repository. Removing this folder restores global behavior.

---

## Global Configuration

Global installation applies the skill to every coding session, regardless of working directory. Use this when you want consistent "action-first, numbered-steps" guidance across all projects.

To install globally:

```bash
npx skills add ayghri/i-have-adhd -g

```

The `-g` flag directs the CLI to copy the skill into your agent's user-wide directory (e.g., `~/.cursor/skills/` or `~/.agents/skills/`).

### Manual Global Installation

For agents not supported by the CLI, copy the skill manually:

```bash

# Cursor-specific global install

mkdir -p ~/.cursor/skills
cp -R i-have-adhd/skills/i-have-adhd ~/.cursor/skills/

# Generic agent global install

mkdir -p ~/.agents/skills
cp -R i-have-adhd/skills/i-have-adhd ~/.agents/skills/

```

The [`README.md`](https://github.com/ayghri/i-have-adhd/blob/main/README.md) (lines 27-33) documents these manual steps and recommends forking the repository, editing [`skills/i-have-adhd/SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md), then swapping in your customized copy.

---

## Choosing Between Project and Global Scope

| Use Case | Recommended Scope | Rationale |
|---------|-----------------|-----------|
| Personal preferences across all codebases | Global | One-time setup, consistent everywhere |
| Team-specific conventions | Project | Version-controlled alongside code |
| Framework-specific adaptations | Project | Tailor steps to project architecture |
| Experimental modifications | Project | Isolate changes, avoid breaking global workflow |

The repository includes [`scripts/run_evals.py`](https://github.com/ayghri/i-have-adhd/blob/main/scripts/run_evals.py), which demonstrates programmatic skill loading via `skill_path.read_text()`—useful for testing scoped configurations.

---

## CI Validation for Scope Consistency

The [`.github/workflows/cursor-skill-sync.yml`](https://github.com/ayghri/i-have-adhd/blob/main/.github/workflows/cursor-skill-sync.yml) workflow ensures project-scoped copies stay synchronized with the canonical skill definition. This prevents drift when teams maintain local modifications.

To implement similar checks in your own repositories:

1. Fork `ayghri/i-have-adhd`
2. Customize [`skills/i-have-adhd/SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md)
3. Install your fork project-wise or globally using `npx skills add <your-fork>`

---

## Summary

- **Project-scoped** skills live in `./skills/` and load only when working in that repository
- **Global** skills live in `~/.<agent>/skills/` and apply to all sessions
- Use `npx skills add ayghri/i-have-adhd` for project installation or add `-g` for global
- Local skills override global ones when both exist
- Modify [`SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/SKILL.md) front-matter in [`skills/i-have-adhd/SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md) to customize invocation behavior

---

## Frequently Asked Questions

### Can I have both global and project-specific versions installed simultaneously?

Yes. When both exist, agents load the local `./skills/` version first, overriding the global configuration for that project only. Remove the local copy to revert to global behavior.

### Where is the skill behavior actually defined?

The core rules and front-matter are in [`skills/i-have-adhd/SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md). This file specifies `name`, `description`, and `disable-model-invocation` settings that control how the agent engages the skill.

### What agents support this installation method?

Claude, Copilot, Cursor, OpenCode, and any agent implementing the *Agent Skills* specification can load skills from both project and global directories. The `npx skills` CLI handles path detection automatically.

### How do I prevent the skill from auto-triggering?

Set `disable-model-invocation: true` in your customized [`SKILL.md`](https://github.com/ayghri/i-have-adhd/blob/main/SKILL.md) front-matter. The agent will then wait for explicit invocation rather than applying the skill automatically to every interaction.