subagent-driven-development

Execute an approved implementation plan in the current session by dispatching one subagent per task, with spec and quality reviews after each task. Use after writing-plans when tasks are mostly independent.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "subagent-driven-development" with this command: npx skills add vibemastery/toolkit/vibemastery-toolkit-subagent-driven-development

VibeMastery Subagent-Driven Development

Implement plan tasks using focused subagents and strict review gates.

Hard Rules

  1. Run only one implementation subagent at a time.
  2. Require two review gates per task:
    • spec compliance review
    • code quality review
  3. Fix all review issues before moving to the next task.
  4. Do not commit changes unless the user explicitly asks.

Workflow

  1. Read the plan once and extract all tasks.
  2. For each task:
    • dispatch implementer subagent with full task text + context
    • answer clarifying questions before implementation continues
    • run task verification checks
    • dispatch spec compliance reviewer
    • if issues: implementer fixes and spec reviewer re-checks
    • dispatch code quality reviewer
    • if issues: implementer fixes and reviewer re-checks
    • mark task complete
  3. After all tasks, run one final end-to-end review.
  4. Present results and request user review.

Prompt Templates

Use these helper templates when dispatching subagents. They are located in the same directory as this skill file (e.g. skills/subagent-driven-development/ or .claude/skills/vibemastery-toolkit/subagent-driven-development/ depending on your install):

  • implementer-prompt.md
  • spec-reviewer-prompt.md
  • code-quality-reviewer-prompt.md

Read the relevant file and use its contents as the subagent prompt, filling in the task text and context placeholders.

Per-Task Output Format

After each task, report:

  • Task completed
  • Files changed
  • Spec review result
  • Quality review result
  • Verification result

Beginner Explanation Standard

  • Keep wording simple and direct.
  • Explain what changed and why it matters.
  • Translate reviewer findings into plain English.

Completion Criteria

This workflow is complete only when:

  1. Every task passes both review gates.
  2. Plan verification checks pass.
  3. Results are explained clearly for beginners.
  4. No commit has been created unless explicitly requested.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

setup-wizard

No summary provided by upstream source.

Repository SourceNeeds Review
General

brainstorming

No summary provided by upstream source.

Repository SourceNeeds Review
General

debug-coach

No summary provided by upstream source.

Repository SourceNeeds Review
General

executing-plans

No summary provided by upstream source.

Repository SourceNeeds Review