subagent-driven-development

What This Skill Does

The subagent-driven-development skill is a high-efficiency framework designed for OpenClaw to execute complex development plans by leveraging specialized, ephemeral subagents. Instead of relying on a single long-lived agent that might suffer from context drift or fatigue, this skill systematically dispatches fresh subagents for every individual task defined in your implementation plan. The process enforces a rigorous, automated two-stage review loop: first, a spec compliance review to ensure the implementation strictly adheres to the requirements, followed by a code quality review to guarantee maintainable, clean code. This modular approach ensures that each component of your codebase is built in isolation, tested, and validated before moving on, resulting in faster development cycles and significantly higher code quality.

Installation

To add this capability to your OpenClaw environment, use the internal skill manager. Open your terminal or command interface and execute the following installation command:

clawhub install openclaw/skills/skills/zlc000190/subagent-driven-development

Ensure that you have appropriate write permissions for your workspace, as the skill requires creating and managing subagent prompts and TodoWrite logs.

Use Cases

This skill is ideal for projects that have a clear implementation roadmap but contain distinct, loosely coupled tasks. Use it when you are working on a multi-feature sprint where you want to maintain focus within the current session without the overhead of manually context-switching between tools. It is particularly powerful for refactoring large modules, adding independent API endpoints, or implementing test suites where granular validation is required for every unit of work.

Example Prompts

1. "I have a 5-step plan for our new authentication service. Initialize subagent-driven-development and start with the database schema task."
2. "We need to implement the three UI components outlined in the project documentation. Use the subagent-driven-development workflow to build them sequentially."
3. "Transition to subagent-driven-development for the remaining items in my current TodoWrite list to ensure strict code quality compliance."

Tips & Limitations

• Preparation is Key: Ensure your initial implementation plan is granular. If tasks are too broad, the subagent will struggle to scope the review process.
• Session Management: This skill works best within a single session. If you need to perform actions across separate infrastructure or environments, consider the 'executing-plans' skill instead.
• Human oversight: While the two-stage review is automated, human intervention remains valuable after the final integrated review step to verify business logic alignment.