Development Roadmap for the i-have-adhd Plugin: 7 Key Milestones Explained

The i-have-adhd plugin roadmap prioritizes expanding runtime support for emerging LLM platforms, enabling user-configurable skill rules without forking, and formalizing release management through a structured ROADMAP.md file.

The i-have-adhd plugin is a lightweight conversational skill designed for coding-assistant runtimes including Claude, Codex, Pi, OMP, OpenCode, Qwen, Kimi, and Gemini. According to the repository structure in ayghri/i-have-adhd, the development roadmap is incremental, test-driven, and focused on widening platform support while giving end-users granular control over the skill’s behavior. The current architecture centers on ten core response rules defined in skills/i-have-adhd/SKILL.md, with future releases building upon this foundation through seven distinct milestones.

Expand Runtime Coverage for Emerging LLM Platforms

The immediate priority is adding or improving adapters for new LLM platforms. The repository currently maintains separate manifests for different runtimes, including .claude-plugin/plugin.json, gemini-extension.json, kimi.plugin.json, and qwen-extension.json.

This modular structure allows maintainers to rapidly integrate support for emerging platforms like Anthropic Claude 3, Google Gemini 1.5, and OpenAI GPT-4-Turbo. The runtime adapters are primarily implemented in:

• hooks/hooks.json – Registers the "always-on" hook that triggers the skill across environments
• extensions/i-have-adhd.ts – TypeScript adapter for Pi and OMP runtimes
• extensions/context-compat.ts – Compatibility layer for varying runtime contexts

Future work involves creating similar adapter files for any new runtime that releases a plugin API, ensuring the skill remains portable across the evolving AI assistant landscape.

Enhance Skill Customisation Without Forking

Currently, users must fork the entire repository to modify the ten core rules defined in skills/i-have-adhd/SKILL.md. The README describes this "Tune it" workflow (lines 81-92) as the primary customization method.

The roadmap includes exposing a configuration layer—likely via a .i-have-adhdrc file or runtime-specific UI—that allows end-users to tweak rules or add new ones without maintaining separate forks. This would transform the static rule set into a dynamic configuration that the runtime reads at load time, making the skill accessible to non-technical users who cannot manage Git workflows.

Improve Internationalisation Beyond Documentation

While the repository already ships with multilingual documentation in .github/readme/ (including README.zh-CN.md and README.pt-BR.md), the skill rules themselves remain English-only.

Upcoming releases will localize the actual behavior of the skill, translating the ten core response rules and adapting test cases for international users. This ensures that the plugin’s ADHD-friendly conversational style—such as "lead with next action" and "number your steps"—functions effectively across languages and cultural contexts.

Increase Test Coverage and CI Robustness

The current CI pipeline in .github/workflows/ includes runtime-load checks (*load-check.yml) and a basic unit-test suite in tests/. The roadmap calls for two specific enhancements:

• Property-based testing for each of the ten rules (e.g., verifying "no preambles" or "max-5-list items" constraints)
• Parallel integration tests covering every runtime adapter simultaneously to prevent platform-specific regressions

These improvements will guard against breaking changes as the skill expands to support additional LLM platforms.

Add a Structured Roadmap File (Issue #127)

The repository currently lacks an explicit ROADMAP.md. According to open issue #127 labeled "AI Agora," maintainers plan to introduce a formal roadmap document that codifies upcoming milestones:

• v2.0: Rule-based templating system
• v2.1: User-configurable time estimates
• v2.2: Community-driven rule extensions

This structured approach will replace the inferred roadmap with explicit version planning and automated changelog generation.

Implement Package Publishing and Semantic Versioning

The presence of package.json and opencode.json indicates readiness for distribution via npm and the OpenCode marketplace. Future releases will adopt semantic versioning with automated changelog generation, moving away from the current installation method that requires manual GitHub references.

Users will eventually install the plugin through standard package managers:

# Future npm distribution (planned)

npm install @ayghri/i-have-adhd

# Current Claude Code installation

claude plugin install i-have-adhd@i-have-adhd

Optimize Performance and Resource Usage

Because the skill runs on-the-fly through an "always-on" hook registered in hooks/hooks.json, the roadmap includes a dedicated performance phase. Key optimizations involve:

• Pre-compiling the rule set to reduce parse time on each invocation
• Caching parsed Zod schemas (if implemented for configuration validation)
• Reducing latency for real-time coding assistant interactions

These changes ensure the plugin remains lightweight even as customization options expand.

Summary

• The i-have-adhd plugin is architected around ten core rules in skills/i-have-adhd/SKILL.md and runtime adapters in extensions/ and hooks/
• The roadmap prioritizes runtime expansion for Claude 3, Gemini 1.5, and GPT-4-Turbo through modular manifest files
• User customization will move from fork-based workflows to configuration files like .i-have-adhdrc
• Internationalisation will extend beyond documentation to localize the actual skill behavior and test suites
• Testing infrastructure will add property-based tests and parallel CI checks for all supported runtimes
• Formal versioning begins with the introduction of ROADMAP.md (issue #127) and semantic release automation
• Performance optimization targets the "always-on" hook architecture to minimize latency during rule execution

Frequently Asked Questions

What is the i-have-adhd plugin used for?

The i-have-adhd plugin is a conversational skill that modifies how coding assistants communicate. According to skills/i-have-adhd/SKILL.md, it enforces ten specific rules such as leading with the next action, numbering steps, and eliminating fluff text. It can be added to any supported runtime including Claude, Codex, OpenCode, and Gemini to make technical explanations more accessible for users who prefer structured, direct communication.

How do I customize the rules without forking the repository?

Currently, you must fork the repository and edit skills/i-have-adhd/SKILL.md directly, then reinstall from your fork using claude plugin install i-have-adhd@<your-username>/i-have-adhd. However, the development roadmap includes plans for a configuration-based system using files like .i-have-adhdrc that would allow rule customization without maintaining a separate fork or Git workflow.

Which files control how the plugin loads in different AI assistants?

Runtime-specific behavior is controlled by multiple files: hooks/hooks.json registers the always-on trigger, extensions/i-have-adhd.ts provides the TypeScript adapter for Pi and OMP runtimes, and platform-specific manifests like .claude-plugin/plugin.json, opencode.json, and gemini-extension.json declare entry points and metadata for each supported assistant. These files collectively ensure the skill loads correctly across Claude, Codex, Gemini, Qwen, and other platforms.

When will the plugin support npm installation?

The repository already contains package.json and opencode.json, indicating technical readiness for npm and marketplace distribution. According to the roadmap, semantic versioning and automated publishing will be implemented alongside the formal ROADMAP.md file (issue #127), likely arriving in the v2.0 release cycle that introduces rule-based templating and community-driven extensions.

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