How to Use i-have-adhd with Zed Editor: Complete Setup Guide

You can use i-have-adhd with Zed by installing its SKILL.md file through Zed's Agent Panel, either via direct URL import from GitHub or by cloning the repository locally to ~/.config/zed/skills/.

The i-have-adhd project provides formatting rules designed for ADHD-friendly readability in AI-assisted coding. Since Zed's built-in Agent natively supports the Skill specification format, integrating this tool requires no conversion or complex configuration—just point Zed to the skill definition and invoke it with a slash command.

---

Two Methods to Install i-have-adhd in Zed

Zed offers two installation paths depending on whether you prefer cloud-based skills or local, offline access. Both methods register the skill in Zed's Agent Panel, making it available across all your projects through the /i-have-adhd command.

Method 1: Install from URL (Recommended)

This approach keeps the skill in Zed's User scope, meaning it's globally available without cluttering individual project directories.

1. Open Zed's Agent Panel and navigate to the Skills manager
2. Select "Create skill from URL" (or press Cmd+Shift+P → type "agent: create skill from url")
3. Paste this exact URL:

https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md

4. Confirm—the skill immediately appears in your User scope

Once installed, typing /i-have-adhd in any file activates the skill's formatting constraints.

Method 2: Local Clone for Offline Access

Copy the skill directly into Zed's configuration directory when you need network-independent operation or want to version-control the skill alongside your own dotfiles.

# Clone the repository

git clone https://github.com/ayghri/i-have-adhd

# Copy skill to Zed's user-skills directory

cp -R i-have-adhd/skills/i-have-adhd ~/.config/zed/skills/

# Refresh Zed's Skills manager or restart the editor

The skill now loads from your local filesystem at ~/.config/zed/skills/i-have-adhd/.

---

How the Zed Integration Works

Understanding the mechanics helps troubleshoot issues and customize behavior.

The SKILL.md Format

Zed's Agent recognizes skills through a SKILL.md file containing:

• A YAML metadata header (--- delimited)
• Rule definitions that constrain LLM output

In skills/i-have-adhd/SKILL.md, the critical disable-model-invocation: true flag instructs Zed not to re-invoke the model when the skill runs. Instead, the Agent injects the pre-written behavioral rules directly into the prompt context.

What Happens at Runtime

When you type /i-have-adhd:

1. Zed locates the registered skill in the Agent Panel
2. Parses the rule definitions from SKILL.md
3. Prepends these constraints to your current LLM prompt
4. The model generates responses following ADHD-friendly patterns:
   • Lead with the immediate next action
   • Number all sequential steps
   • Provide concrete time estimates
   • Break complex explanations into scannable chunks

---

Verifying and Managing Your Installation

Check Installation Status

Open the Skills manager in Zed's Agent Panel and look for "i-have-adhd" in the list, or simply type /i-have-adhd in any editor buffer—the autocomplete should surface the skill if properly registered.

Update the Skill

Keep current with upstream improvements:

URL-based installations: Repeat the "Create skill from URL" step with the same GitHub URL—Zed overwrites the previous version.

Local installations:

cd i-have-adhd && git pull
cp -R skills/i-have-adhd ~/.config/zed/skills/

# Restart Zed or reload the Skills manager

---

Key Files and References

File	Location	Purpose
SKILL.md	skills/i-have-adhd/SKILL.md	Core skill definition consumed by Zed's Agent; contains metadata header and formatting rules
INSTALL.md	Repository root	Official installation guide with Zed-specific instructions in the dedicated Zed section
README.md	Repository root	Project overview with quick-start examples across multiple editors and agents

The Zed integration follows the standard documented in INSTALL.md under the Zed heading, which mirrors the procedures outlined above.

---

Summary

• Zed natively supports the i-have-adhd skill through its Agent Panel—no plugins or conversions needed
• Two installation methods: URL import for convenience (https://github.com/ayghri/i-have-adhd/blob/main/skills/i-have-adhd/SKILL.md), or local clone to ~/.config/zed/skills/ for offline use
• Activation: Type /i-have-adhd in any file to apply ADHD-friendly formatting constraints to LLM responses
• Updates: Re-import via URL or git pull and recopy for local installations

---

Frequently Asked Questions

What is the i-have-adhd skill for Zed?

The i-have-adhd skill is a set of LLM prompt constraints that restructure AI-generated code explanations for ADHD-friendly readability. According to the ayghri/i-have-adhd source code, it enforces patterns like numbered steps, action-first statements, and concrete time estimates. In Zed, this skill integrates directly with the built-in Agent through the standard Skill specification format.

Where does Zed store locally installed skills?

Zed loads user-scoped skills from ~/.config/zed/skills/ on macOS and Linux. When you copy the i-have-adhd directory there—as shown in the local installation method—the Agent discovers it on startup without requiring network access to GitHub.

Can I use i-have-adhd without installing it permanently?

No temporary or per-session option exists in Zed's current Agent implementation. You must register the skill through either URL import or local filesystem placement before the /i-have-adhd slash command becomes available. However, once registered, you choose when to invoke it—the skill doesn't automatically apply to all Agent interactions.

Does the i-have-adhd skill work with all Zed LLM providers?

Yes. The skill operates at the prompt-construction layer before your request reaches any LLM backend. The disable-model-invocation: true flag in skills/i-have-adhd/SKILL.md ensures Zed only manipulates the prompt context rather than triggering model-specific behavior, making it provider-agnostic across OpenAI, Anthropic, and local model configurations.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}