A high-velocity execution framework that dispatches specialized subagents for independent tasks with automated two-stage reviews.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install subagent-driven-development
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install subagent-driven-development using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Subagent-Driven Development?
Subagent-Driven Development is a sophisticated workflow designed for AI coding agents to execute implementation plans within a single session. By leveraging the modular power of Openclaw Skills, this method dispatches a fresh subagent for every task to prevent context pollution and ensure high-speed iteration.
The core philosophy centers on a Fresh Subagent per Task model followed by a rigorous two-stage review process. This ensures that every piece of code not only meets the initial specifications but also adheres to high-quality standards before moving to the next task, making it a premier choice among Openclaw Skills for complex software engineering.
Subagent-Driven Development Use Cases
- Executing complex implementation plans with multiple independent tasks within a single session.
- Maintaining high code quality through automated spec compliance and quality reviews.
- Reducing context switch overhead by staying within a single agent session instead of switching to parallel sessions.
- Scaling development velocity using Openclaw Skills for systematic task dispatching.
How Subagent-Driven Development Works
- Read the implementation plan and extract all tasks with full text and relevant context.
- Create a TodoWrite list to track the status of each task throughout the session.
- Dispatch an Implementer subagent for the current task to handle implementation, testing, and initial self-review.
- Trigger a Spec Reviewer subagent to verify that the implementation matches the original requirements exactly.
- Invoke a Code Quality Reviewer subagent to approve the technical implementation, style, and efficiency.
- Repeat the cycle for all remaining tasks, ensuring every review loop is closed before proceeding.
- Perform a final code review for the entire implementation using integrated Openclaw Skills.
Subagent-Driven Development Setup
To utilize this skill, you must have an implementation plan ready. The process integrates with other Openclaw Skills to maintain isolation and quality.
# Ensure your environment is prepared with the following prompt templates:
# - ./implementer-prompt.md
# - ./spec-reviewer-prompt.md
# - ./code-quality-reviewer-prompt.md
# Initialize an isolated workspace before starting
# This usually involves using git worktrees for clean environments.
Subagent-Driven Development Data Schema & Taxonomy
The skill manages task execution and verification through a structured data lifecycle:
| Component | Purpose | Data Managed |
|---|---|---|
| TodoWrite | Task Tracking | Stores full task text, context, and completion status. |
| Implementer | Task Execution | Generates code, tests, and commit messages. |
| Spec Reviewer | Compliance | Validates code against documentation and feature requirements. |
| Quality Reviewer | Optimization | Evaluates code for magic numbers, complexity, and best practices. |
All data is passed via specialized prompts within the Openclaw Skills framework to ensure zero context drift.
Subagent-Driven Development Advanced Features
- Automated two-stage review loops where spec compliance must be achieved before quality review starts.
- Context-isolated task execution to prevent hallucination or context pollution in long-running sessions.
- Native Test-Driven Development (TDD) enforcement for all implementation subagents.
- Direct integration with Openclaw Skills like git-worktrees for safe, branch-based development.
- Recursive review loops that automatically trigger fixes until both spec and quality gates are passed.
SKILL.md
Loading
Related Openclaw Skills
Requesting Code Review
A specialized skill for triggering automated subagent code reviews to ensure quality and requirement alignment during development.
Finishing a Development Branch
A structured workflow for finalizing development work, ensuring all tests pass before automating merges or pull request creation.
Dispatching Parallel Agents
A sophisticated orchestration pattern for running multiple AI agents concurrently to solve independent technical problems without shared state interference.
Brainstorming
A collaborative design skill that refines project ideas into structured technical specifications through iterative dialogue.
Using Git Worktrees
A specialized skill for creating and managing isolated Git workspaces with automated safety checks and environment setup.
Using Superpowers: AI Skill Orchestration
A mandatory meta-skill that enforces the invocation of relevant specialized skills before an AI agent takes any action or answers any question.
Featured*
