subagent-driven-development

Dispatch isolated subagents to implement, test, and review tasks.

Updated Jan 25, 2026
One-click install
npx skills add https://github.com/MINT-IA/MINT --skill subagent-driven-development-mint-ia
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/MINT-IA/MINT/tree/main/.claude/skills/subagent-driven-development
Command: npx skills add https://github.com/MINT-IA/MINT --skill subagent-driven-development-mint-ia

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coordinating implementation work across multiple independent tasks often leads to context pollution, inconsistent quality, and slow iteration when a single agent or human must manage every step; this Skill enforces isolated execution and structured reviews so tasks complete reliably and quickly.

Core Features & Use Cases

  • Isolated subagents per task: Dispatch a fresh implementer subagent for each task so context is precise and contained.
  • Two-stage review pipeline: Enforce spec compliance review first, then a separate code-quality review to catch both requirement mismatches and maintainability issues.
  • Model and status guidance: Guidance for choosing models by task complexity and handling implementer statuses (DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, BLOCKED).
  • Use case: Execute a multi-task feature plan in-session, with each task implemented, tested, self-reviewed, spec-reviewed, and quality-reviewed before marking complete.

Quick Start

Dispatch subagents to implement tasks from my plan, answer any subagent questions, then run spec compliance followed by code-quality review for each task.

Frequently Asked Questions about subagent-driven-development

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I execute multiple independent development tasks without context pollution?

To execute independent development tasks without context pollution, dispatch a fresh, isolated subagent per task to implement, test, and commit work. This ensures context remains precise and contained, preventing cross-task interference and maintaining consistent quality.

What is the best way to automate spec compliance and code quality reviews?

Automate spec compliance and code quality reviews using a two-stage review pipeline. Enforce a spec compliance review first to catch requirement mismatches, followed by a separate code-quality review to identify maintainability issues before marking the task complete.

How do I orchestrate subagents for task implementation and testing?

Orchestrate subagents for task implementation by dispatching isolated workers per task. Provide fresh task context, select models based on task complexity, and handle statuses like DONE, NEEDS_CONTEXT, or BLOCKED to ensure reliable and fast iteration.

Can I run a TDD workflow using isolated subagents in a single session?

Yes, you can run a TDD workflow in a single session by dispatching isolated subagents to implement, test, and self-review tasks. Execute mostly independent tasks from your plan sequentially, ensuring each passes spec and code-quality reviews before completion.

When should I not use subagent-driven development for task orchestration?

Avoid subagent-driven development for tightly coupled tasks requiring continuous shared context. This approach is designed for mostly independent tasks where isolated execution, fresh context, and structured two-stage reviews provide the most benefit.

How do I handle a subagent that returns a BLOCKED or NEEDS_CONTEXT status?

Handle BLOCKED or NEEDS_CONTEXT statuses by answering the subagent's questions or providing the missing task context. The framework includes status handling to pause execution, request information, and resume once the implementer has what it needs.