subagent-driven-development

Executes implementation plans by dispatching fresh subagents per task with per-task and final code reviews.

10|1|Updated Jul 29, 2026
One-click install
npx skills add https://github.com/flowing-abyss/obsidian-local-fonts --skill subagent-driven-development-flowing-abyss
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/flowing-abyss/obsidian-local-fonts/tree/main/.ai/skills/subagent-driven-development
Command: npx skills add https://github.com/flowing-abyss/obsidian-local-fonts --skill subagent-driven-development-flowing-abyss

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Executing a multi-task implementation plan in one AI session pollutes context, loses progress after compaction, and ships unreviewed code. This Skill orchestrates plan execution through isolated subagents, persistent ledgers, and enforced review gates so every task is implemented, tested, and verified. ## Core Features & Use Cases - Fresh subagent per task: Dispatches an implementer subagent with a task brief file, keeping the controller's context clean for coordination. - Two-stage review loop: Runs a spec-compliance and code-quality review after each task, plus a broad whole-branch review at the end, with a bounded five-round fix loop and adjudication rules. - Crash-safe progress ledger: Tracks task completion, fix rounds, and parked findings in a per-plan workspace so work resumes correctly after context compaction. - Use Case: You have a written implementation plan with eight independent tasks. This Skill dispatches an implementer for each task, reviews every diff, escalates stuck tasks to more capable models, and finishes with a whole-branch review before merging. ## Quick Start Use the subagent-driven-development skill to execute the implementation plan at docs/plans/feature-plan.md task by task with reviews.

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 an implementation plan with AI subagents?

Dispatch a fresh implementer subagent per task with a task brief file, then run a spec-compliance and code-quality review on each task's diff. Track progress in a ledger file and finish with a whole-branch review before merging.

What is the difference between subagent-driven development and executing plans directly?

Subagent-driven development stays in the same session but gives each task a fresh subagent with isolated context, avoiding context pollution. Direct plan execution typically runs in a parallel session with human check-ins between tasks.

How does the skill recover progress after context compaction?

It writes a per-plan ledger at .superpowers/sdd/<plan>/progress.md recording completed tasks, commits, and fix rounds. After compaction, the controller trusts the ledger and git log rather than its own memory.

When should I not use subagent-driven development?

Avoid it when tasks are tightly coupled and cannot be implemented independently, or when no implementation plan exists yet. In those cases, brainstorm or plan first, or execute manually.

How many fix rounds are allowed per task review?

A maximum of five fix rounds per task. Rounds 1-3 resume the original implementer, rounds 4-5 dispatch a fresh implementer on a more capable model, and remaining findings are adjudicated or escalated as BLOCKED.