adr-gen

What This Skill Does

The adr-gen skill is an essential tool for engineering teams aiming to maintain a high-quality, searchable, and historical knowledge base of technical decisions. By leveraging AI to draft Architecture Decision Records (ADRs), this skill ensures that significant technical choices—such as selecting a database engine, adopting a new framework, or changing communication protocols—are documented consistently. Instead of losing vital context to memory or team turnover, adr-gen helps you create structured documentation that outlines the problem, the considered alternatives, the rationale behind the final decision, and the long-term trade-offs. It transforms a simple prompt into a professional, industry-standard ADR document, ensuring future team members understand the 'why' behind the codebase.

Installation

You can install this skill directly through your OpenClaw terminal interface. Simply execute the following command:

clawhub install openclaw/skills/skills/lxgicstudios/adr-gen

Once installed, the tool is immediately accessible as an npx-compatible executable, requiring no further configuration or API keys to function on the free tier.

Use Cases

• Documentation Automation: Quickly generate records after sprint planning sessions where major architectural shifts were agreed upon.
• Onboarding Assistance: Provide new developers with a clear history of why existing infrastructure exists, reducing the burden on senior engineers.
• Architecture Review Prep: Generate formal documentation required for high-stakes technical reviews or security audits.
• Technical Debt Management: Keep track of temporary workarounds that were chosen as compromises for current deadlines.

Example Prompts

• "Create an ADR for our decision to switch from React state to Zustand for global state management to reduce boilerplate."
• "Generate an ADR for implementing Redis as a caching layer. Context: We are seeing high latency on our profile endpoints and need to reduce database load."
• "Document the choice to use TypeScript in our new microservices repo, detailing why we chose it over keeping the legacy JavaScript stack."

Tips & Limitations

• Be Specific: The AI performs best when you provide constraints. Include technical details like 'high concurrency' or 'limited budget' in your prompt.
• Review Mandatory: While the AI generates excellent drafts, ensure that a human architect verifies the consequences section, as only you know your team's specific long-term maintenance capacity.
• File Management: Ensure you specify the --output flag if you want to save the generated markdown directly into your existing project repository structure.