ss-subagent-driven-development

Dispatch subagents for per-task implementation with two-stage reviews.

2|2|Updated Mar 25, 2026
One-click install
npx skills add https://github.com/CoyoteLeo/superspec --skill ss-subagent-driven-development
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ss-subagent-driven-development
Source: https://github.com/CoyoteLeo/superspec/tree/main/skills/ss-subagent-driven-development
Command: npx skills add https://github.com/CoyoteLeo/superspec --skill ss-subagent-driven-development

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrate per-task execution of implementation plans by dispatching dedicated subagents for each task, with a two-stage review after every task to ensure spec compliance and code quality, while maintaining both in-session and persistent task tracking.

Core Features & Use Cases

  • Fresh subagent per task with isolated context, reducing context leakage and enabling focused work.
  • Two-stage review after each task: spec compliance review followed by code quality review.
  • Per-task and per-change tracking using TodoWrite (in-session) and tasks.md (persistent) to resume work across conversations.
  • Inline mode option for direct execution when subagents are unavailable or tightly coupled tasks.
  • Orchestrate end-to-end work from a single plan by dispatching implementer, spec-reviewer, and code-reviewer subagents per task.

Quick Start

Load the plan from changes/YYYY-MM-DD-<topic>/plan.md and dispatch the first implementer subagent to execute its tasks.

Frequently Asked Questions about ss-subagent-driven-development

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

FAQPage Schema
How do I execute implementation plans with isolated context per task?

Plan execution with isolated context uses dedicated subagents dispatched per task, preventing context leakage. Each subagent handles its specific task independently, and a two-stage review validates spec compliance and code quality before moving to the next task.

What is two-stage review in subagent-driven task orchestration?

Two-stage review in task orchestration applies a spec compliance review followed by a code quality review after each task. This mechanism ensures the subagent's implementation meets the plan's requirements before proceeding to subsequent tasks.

How do I track implementation progress across conversations and sessions?

Progress tracking across conversations uses synchronized in-session TodoWrite and persistent tasks.md files. This dual tracking records per-task and per-change status, allowing you to resume plan execution from the exact point of interruption in a new session.

Can I execute tasks inline instead of dispatching subagents?

Yes, inline mode enables direct task execution instead of dispatching subagents. You use inline mode when subagents are unavailable or when handling tightly coupled tasks that require shared context within the implementation plan.

What directory structure is required for subagent-driven plan execution?

Subagent-driven plan execution requires a changes/YYYY-MM-DD-<topic>/ directory structure. You load the implementation plan from the plan.md file within this directory, and the orchestration tracks progress relative to this change-specific path.

When should I not use fresh subagents for plan execution?

You should avoid fresh subagents for plan execution when tasks are tightly coupled and require shared context, or when subagents are unavailable in your environment. In these cases, switch to inline mode for direct task processing.