Wan Huiyan Subagent Driven Development
- 1 repo stars
- Updated April 15, 2026
- wan-huiyan/subagent-driven-development
wan-huiyan-subagent-driven-development is a Claude Code plugin marketplace that installs a skill for developers who need fresh subagents per implementation task with two-stage spec and code-quality review gates plus opti
About
wan-huiyan-subagent-driven-development is a Claude Code plugin marketplace (version 1.0.0) distributing a skill adapted from obra/superpowers subagent-driven-development with added parallel worktree mode from spawn-tasks. Given an implementation plan, the controller extracts tasks, analyzes file dependencies, and dispatches a fresh implementer subagent per task to avoid context pollution. Each task passes two review gates in order: a spec compliance reviewer verifies requirements were met, then a code quality reviewer checks cleanliness and tests—loops repeat until approved. Independent tasks can run in parallel with git worktree isolation after user confirmation, followed by a cross-task integration review. Three prompt templates define implementer, spec-reviewer, and code-quality-reviewer behavior. Reach for this skill when executing multi-task plans where unchecked drift across tasks would accumulate errors—not for tightly coupled single-file edits.
- Dispatches fresh implementer subagent per task to prevent context pollution
- Two-stage review: spec compliance first, then code quality, loop until approved
- Parallel mode runs independent tasks in git worktrees with cross-task integration review
- Includes 3 prompt templates: implementer, spec-reviewer, code-quality-reviewer
- Auto-detects sequential vs parallel mode from task dependency analysis
Wan Huiyan Subagent Driven Development by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add wan-huiyan/subagent-driven-developmentAdd your badge
Show developers this marketplace is listed on Skillselion. Paste this into your README.
| repo stars | ★ 1 |
|---|---|
| Last updated | April 15, 2026 |
| Repository | wan-huiyan/subagent-driven-development ↗ |
How do you execute multi-task plans with subagent review gates?
Hand Claude Code a 5-task implementation plan and invoke Subagent-Driven Development to dispatch fresh subagents with spec-then-quality review gates per task.
Who is it for?
Developers executing multi-task implementation plans in Claude Code who want mandatory spec-then-quality review gates and optional parallel worktree isolation for independent tasks.
Skip if: Single trivial edits or tightly coupled tasks sharing files where sequential-only mode still risks merge conflicts and review overhead exceeds the task value.
What you get
Completed tasks with passed spec and code-quality reviews, git commits per task, and cross-task integration review after parallel runs.
- reviewed task commits
- spec compliance reports
- integration review summary
By the numbers
- Version 1.0.0 with 3 subagent prompt templates
- Two mandatory review gates per task: spec compliance then code quality
Plugins in this marketplace
1 plugin - install individually after you add the marketplace.
Recommended Marketplaces
How it compares
Choose subagent-driven-development over fire-and-forget parallel spawning when every task must pass spec and quality review loops, including post-parallel integration checks.
FAQ
Why does subagent-driven-development use two review stages?
Subagent-driven-development checks spec compliance first—did the implementer build what was asked—then code quality second. A well-written function solving the wrong problem still fails, so spec review prevents wasted effort reviewing code that must be rewritten.
When does parallel mode activate?
Parallel mode in subagent-driven-development activates when dependency analysis shows tasks touch completely different modules, after explicit user confirmation. Tasks sharing files run sequentially; after parallel completion, one cross-task integration reviewer checks all change