How to Use i-have-adhd with OpenCode: A Complete Setup Guide

To use i-have-adhd with OpenCode, install the skill via the CLI and activate it with the /i-have-adhd slash command to reshape LLM outputs into action-first, numbered steps optimized for ADHD readers.

The i-have-adhd repository by ayghri provides a specialized skill (plugin) for the OpenCode coding assistant that reformats AI responses for better focus and clarity. By installing this lightweight plugin, you transform verbose assistant outputs into concise, executable instructions that follow ADHD-friendly cognitive patterns.

What is the i-have-adhd Skill?

The i-have-adhd skill is a declarative plugin designed specifically for OpenCode that intercepts and rewrites LLM responses according to ten strict formatting rules. Unlike complex integrations, this skill contains no runtime code; it operates purely through OpenCode's plugin framework using configuration files stored in skills/i-have-adhd/.

The skill enforces action-oriented, time-boxed communication by requiring every response to lead with an immediate next step, number multi-step sequences, suppress tangential explanations, and restate the current state each turn. These rules are defined in skills/i-have-adhd/SKILL.md, which serves as the central instruction set that OpenCode applies to every output when the skill is active.

How to Install i-have-adhd in OpenCode

Installation requires adding the skill to OpenCode's plugin directory or installing directly via the marketplace interface.

Run the following command to install the skill using the OpenCode CLI:

opencode plugin install i-have-adhd@i-have-adhd

This command registers the plugin metadata from plugin.json and copies the skill files into your OpenCode environment. No external API keys or service configurations are required, as the skill operates entirely within OpenCode's local plugin framework.

Activating and Using the Skill

Once installed, activate the skill for your current session using the designated slash command.

Type the following in your OpenCode chat interface:

/i-have-adhd

The skill remains active for the entire session until explicitly disabled. To return to standard output mode, say:

stop adhd mode

Alternatively, you can use "normal mode" to deactivate the skill and restore default response formatting.

How the Skill Transforms LLM Output

When active, the skill rewrites every assistant response according to the rule set defined in SKILL.md. The transformation follows a strict structure:

• Lead with the next action: The first line must be an executable command or immediate step.
• Number multi-step tasks: Complex workflows appear as numbered lists (1, 2, 3).
• Suppress tangents: Explanations are limited to essential context only.
• Restate state each turn: Each response briefly recaps what was just completed.
• Provide concrete next actions: Every response ends with a clear, singular next step.

Example of transformed output:

Run `npm install jsonwebtoken`, then edit src/auth.ts:42.
1. Open src/auth.ts
2. Replace verifyToken (lines 42-58) with the snippet below
3. Run npm test -- auth.spec.ts
Next: run `npm test` and paste the first failing line.

This format eliminates decision paralysis by presenting exactly one path forward with explicit file paths and line numbers.

Key Files and Architecture

The repository uses a lightweight, declarative architecture with no runtime dependencies. Understanding these core files helps with customization:

File	Purpose
skills/i-have-adhd/SKILL.md	Contains the complete rule set (ten formatting rules) that OpenCode uses to reshape outputs.
plugin.json	Plugin metadata required by OpenCode to register and identify the skill at load time.
agents/openai.yaml	Example agent configuration for OpenAI models, showing how to bind the skill rules to specific LLM providers.
agents/gemini.toml	Example agent configuration for Gemini models, demonstrating cross-provider compatibility.

Because the skill is purely declarative, you can modify SKILL.md to adjust formatting rules without changing any code. The plugin framework handles the interception and rewriting automatically based on these definitions.

Summary

• i-have-adhd is a declarative OpenCode skill that reformats AI responses for ADHD-friendly readability according to ten specific rules.
• Install using opencode plugin install i-have-adhd@i-have-adhd and activate with /i-have-adhd.
• The skill enforces action-first formatting, numbered steps, and suppressed tangents in every response.
• Deactivate anytime by saying "stop adhd mode" or "normal mode".
• Core files include SKILL.md (rules), plugin.json (metadata), and optional agent configs in agents/.

Frequently Asked Questions

How do I customize the formatting rules for i-have-adhd?

Edit the skills/i-have-adhd/SKILL.md file in your OpenCode plugins directory. This markdown file contains the ten declarative rules that govern output formatting; modifying these instructions immediately changes how OpenCode reshapes responses without requiring a restart or code compilation.

Does i-have-adhd work with all OpenCode LLM providers?

Yes. The skill includes example configurations for both OpenAI (agents/openai.yaml) and Gemini (agents/gemini.toml) models, and the declarative rule structure is provider-agnostic. The skill operates at the output formatting layer, meaning it works with any underlying LLM that OpenCode supports.

Can I use i-have-adhd with other OpenCode skills simultaneously?

Yes. The skill runs as a session-level output filter alongside other OpenCode plugins. However, ensure that other active skills do not conflict with the formatting constraints defined in SKILL.md, particularly regarding response length and structure, as competing formatting rules may produce inconsistent output.

What happens if I forget to deactivate the skill?

The skill remains active only for the duration of your current OpenCode session. When you start a new session or restart the application, you will need to re-invoke /i-have-adhd to reactivate the ADHD-friendly formatting. The skill does not persist across sessions by default.

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