skills.rest
HumanBean17
HumanBean17COMMUNITY
Explore all skills by this author →
GitHub

subagent-driven-development

Dispatch isolated subagents to implement and review tasks from a structured plan file.

8|1|Updated Apr 20, 2026
Search Tags:#automation#workflow#tdd#development#code review#subagent
One-click install
npx skills add https://github.com/HumanBean17/jrag --skill subagent-driven-development-humanbean17
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/HumanBean17/jrag/tree/main/.claude/skills/subagent-driven-development
Command: npx skills add https://github.com/HumanBean17/jrag --skill subagent-driven-development-humanbean17

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill solves the problem of context pollution and cognitive overload during long-running development tasks by delegating specific implementation steps to isolated, specialized subagents.

Core Features & Use Cases

  • Isolated Execution: Dispatches fresh subagents for each task, ensuring they only have the context necessary for their specific job.
  • Automated Quality Gates: Enforces a strict review loop for every task, requiring both spec compliance and code quality verification before marking a task as complete.
  • Durable Progress Tracking: Maintains a ledger of completed tasks to allow for seamless recovery and resumption of work even after session compaction.

Quick Start

Use the subagent-driven-development skill to execute the implementation plan located at docs/superpowers/plans/active/feature-plan.md by creating a todo list and dispatching the first implementer subagent.

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 complex development plans without context pollution in long coding sessions?▼

You can execute complex development plans by dispatching isolated subagents for each specific task, ensuring they only receive the context necessary for their job. This prevents cognitive overload during long-running development workflows.

How do I automate code review and spec compliance for multi-task development workflows?▼

Automated code review is enforced through a strict review loop requiring both spec compliance and code quality verification before marking any task complete. Subagents independently handle implementation and review for iterative progress.

What do I need to set up autonomous subagents for software engineering tasks?▼

You need git for version control and a structured plan file to define tasks, global constraints, and acceptance criteria. The plan file acts as the source of truth for dispatching independent implementation and review subagents.

How does durable progress tracking work for development tasks after session compaction?▼

Durable progress tracking maintains a ledger of completed tasks, allowing seamless recovery and resumption of work even after session compaction. This ensures no progress is lost when context windows are cleared.

Can I use subagent-driven development for TDD workflows with strict acceptance criteria?▼

Yes, subagent-driven development applies to complex software engineering workflows requiring high-quality, iterative progress with strict spec compliance. It dispatches fresh subagents to handle TDD implementation and review cycles.

When should I avoid using autonomous subagents for development plans?▼

You should avoid using autonomous subagents for simple, single-step tasks that do not require isolated execution or automated quality gates. The workflow is designed for complex, multi-task plans needing strict spec compliance and iterative review.

Related Skills

Explore More Software Engineering Skills→
DYAI2025DYAI2025
3

using-git-worktrees

Create isolated git worktrees and run project install, build, and test commands.

Community
Intermediate
NSObjectsNSObjects
4

requesting-code-review

Dispatch specpowers:code-reviewer subagent and synthesize specialist findings into one review.

Community
Advanced
bobmatnycbobmatnyc
146

Requesting Code Review

Dispatch a code-reviewer subagent to review changes against plans and requirements.

Community
Basic
CodingCossackCodingCossack
19

subagent-driven-development

Executes sequential subagents with two-stage reviews for multi-task plans.

Community
Advanced