plan-i

What This Skill Does

The plan-i (Plan Initializer) skill is a critical component of the OpenClaw AI ecosystem, designed specifically to bootstrap new development or research projects. Its primary function is to transform vague user intent into a structured, file-backed planning document. By enforcing a naming convention and an automated initialization process, it ensures that every project starts with a clean, well-defined foundation, preventing messy file structures and helping the AI agent maintain context across complex tasks. It acts as the gatekeeper for new initiatives, ensuring that analysis precedes implementation.

Installation

To integrate this skill into your environment, run the following command in your terminal: clawhub install openclaw/skills/skills/9talk/plan-i

Use Cases

• Project Kickoff: Use this when you are ready to start a new feature, module, or architecture refactor and need to formalize your requirements.
• Problem Solving: Use it to document issues that require deep investigation before any coding work begins.
• Architectural Design: Ideal for mapping out data models, integration points, and high-level logic flows before touching existing codebases.

Example Prompts

1. "I need to build a new authentication module using OAuth2 for our dashboard."
2. "We need to implement a data sync worker that handles conflict resolution for offline edits."
3. "Start a new plan to refactor the payment gateway integration for the mobile app."

Tips & Limitations

• Naming Matters: The skill strictly enforces slug-style naming (e.g., auth-module-refactor). Use descriptive, hyphenated names.
• No Coding Allowed: plan-i is strictly for planning. It is not designed for coding, debugging, or generating implementation code. If you attempt to use it for coding, the agent will guide you back to the planning phase.
• The Handover: Always follow the agent's prompt to trigger /plan-c after the initial setup. This sequence is vital for the agent to read your newly created documentation and begin a deep-dive analysis of your codebase. Never attempt to manually edit the plan files without triggering the proper agent workflow, as this ensures consistency in state management.