A sophisticated workflow that dispatches fresh subagents per task to ensure high-quality, parallel-safe execution within a single development session.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install subagent-driven-development-3
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-3 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 high-integrity methodology within the Openclaw Skills ecosystem designed to execute implementation plans by isolating tasks. By dispatching a fresh subagent for every individual task, the system effectively eliminates context pollution and ensures that each unit of work is handled with a clean slate. This approach is superior to generic plan execution because it integrates multi-stage review gates, including domain-specialist routing and independent test verification.
This skill is particularly effective for developers who need to stay within a single session while maintaining the rigor of a multi-agent workflow. It leverages specialized agents for backend, frontend, and DevOps tasks, ensuring that every line of code is reviewed by an agent with the appropriate expertise. Through the use of Openclaw Skills, this process automates the transition from implementation to verification, resulting in faster iteration cycles and higher code reliability.
Subagent-Driven Development Use Cases
- When executing an implementation plan where tasks are mostly independent.
- Scenarios requiring specialized domain knowledge for code reviews (e.g., security, accessibility, or backend architecture).
- Situations where you want to avoid context switch overhead by staying in the current session.
- Development environments where Test-Driven Development (TDD) and independent verification are mandatory.
How Subagent-Driven Development Works
- The controller reads the provided implementation plan and extracts all tasks with their full context.
- A fresh implementer subagent is dispatched for the current task using specialized prompt templates.
- The implementer performs the work, runs tests, and commits the code after a self-review.
- An independent verification subagent is invoked to confirm that the test results are accurate and evidence-based.
- The task is routed to a domain-specialist reviewer based on keywords to ensure spec compliance.
- A code quality reviewer provides a final check for architectural standards and best practices.
- Once all gates are passed, the task is marked complete in the tracking system, and the next task begins.
Subagent-Driven Development Setup
To utilize this skill, you must have your environment configured with the necessary prompt templates. Ensure the following files are available in your local repository:
# Core implementation and review templates
touch implementer-prompt.md
touch spec-reviewer-prompt.md
touch code-quality-reviewer-prompt.md
# Advanced verification and recovery templates
touch verification-prompt.md
touch fix-subagent-prompt.md
You should also define your specialist routing logic in a reference file to allow Openclaw Skills to correctly identify and dispatch domain-specific agents for backend, frontend, or infrastructure tasks.
Subagent-Driven Development Data Schema & Taxonomy
The workflow maintains a structured state for every execution cycle to ensure transparency and recoverability:
| Data Component | Purpose |
|---|---|
| TodoWrite | A persistent list tracking task statuses (Pending, In-Progress, Done). |
| Task Context | A snapshot of the specific requirements and environment for the current subagent. |
| Review Logs | Documentation of feedback from verification, spec, and quality subagents. |
| Git SHAs | Unique identifiers for implementation and fix commits created during the session. |
Subagent-Driven Development Advanced Features
- Domain-Specialist Routing: Automatically detects task types (e.g., API, UI, DevOps) and dispatches specialized Openclaw Skills for review.
- Independent Test Verification: A mandatory gate where a separate agent validates test passing claims to prevent implementer bias.
- Systematic Failure Recovery: Integrates with debugging protocols to investigate root causes and dispatch targeted fix subagents when a task fails.
- Multi-Stage Review Loops: Support for optional third-party review gates like security audits or accessibility checks for sensitive features.
- Context Isolation: Ensures fresh subagent instances per task to prevent historical context from polluting new implementation attempts.
SKILL.md
Loading
Related Openclaw Skills
Telegraph Backlinks
Automated publishing tool for creating high-authority, dofollow backlink articles on Telegraph without manual account setup.
Nansi
A mobile-first, no-code AI website builder that creates professional websites and web apps through WhatsApp chat and voice notes.
LinkdAPI Python SDK
A high-performance Python SDK for programmatically accessing LinkedIn profile, company, and job data with enterprise-grade reliability.
StashDog MCP Inventory Skill
An MCP-powered inventory management skill that allows AI agents to list, search, and organize physical or digital items via the StashDog API.
Project Code Standard
A comprehensive automation skill for checking, fixing, and enforcing code standards across Python, JavaScript, and TypeScript projects.
AKShare Data Router (CN)
A sophisticated data routing skill that maps Chinese financial market queries to specific AKShare data sources and analytical processing recipes.
Featured*
