subagent-dev

What This Skill Does

The subagent-dev skill is a sophisticated implementation framework designed to optimize the execution of complex coding tasks within the OpenClaw ecosystem. It functions by orchestrating independent subagents for each task in a development plan, enforcing a rigorous two-stage review process. By decoupling the 'What' (specification compliance) from the 'How' (code quality), this skill ensures that large-scale development projects remain manageable, bug-free, and aligned with initial project requirements. It acts as an extension to the established ralph-loop methodology, providing a granular execution layer that prevents cognitive overload and context leakage in long-running development sessions.

Installation

To install this skill, run the following command in your terminal: clawhub install openclaw/skills/skills/kjaylee/subagent-dev

Use Cases

• Complex Feature Development: Best for large tasks that require multiple steps, ensuring each component meets strict criteria.
• Refactoring Projects: Ideal for systematic code updates where specific functionality needs to be preserved while improving internal quality.
• High-Security Applications: Where rigorous validation of requirements and code patterns is mandatory to prevent security vulnerabilities.
• Team-based Automation: Useful for ensuring consistent development standards across different parts of a codebase.

Example Prompts

1. "Apply the subagent-dev skill to execute the current plan for implementing the User Authentication module. Start by breaking down the tasks and assigning the first subagent."
2. "I need to refactor the database connector. Use subagent-dev to ensure the implementation stays within the defined spec and passes the quality review gate."
3. "Execute the next task in our current plan using subagent-dev, focusing specifically on adding the required unit tests for the payment gateway."

Tips & Limitations

• Tip: Ensure your input specifications are highly detailed, as the subagent's performance in the first stage of the review depends entirely on the clarity of the provided requirements.
• Tip: Use this skill for critical paths; due to the three-tier agent usage (implementer, spec reviewer, quality reviewer), it involves higher API token consumption than single-agent execution.
• Limitation: The skill requires an existing, well-defined plan. It is not designed for open-ended exploratory coding.
• Limitation: As an agent-based orchestration tool, ensure your environment has sufficient permissions for code execution and file modifications, as the agents will automatically write and test code.