Subagent Driven Development

Here's a workflow for executing implementation plans by dispatching fresh Claude instances per task, then running two-stage reviews on each result. You extract all tasks from your plan, dispatch an implementer subagent with isolated context, then run spec compliance and code quality reviews before moving to the next task. The key insight is using fresh context per task to avoid pollution while staying in your current session for coordination. Works well when you have independent tasks that don't need tight coupling. Includes prompt templates for implementer, spec reviewer, and quality reviewer roles, plus logic for handling different completion statuses and selecting appropriate model tiers based on task complexity.

install copies · 90 days. Times the install command was copied on this site in the last 90 days, bots filtered.: 31
skills.sh installs: 252K
GitHub stars: 315.2K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add obra/superpowers --skill subagent-driven-development --agent claude-code

Installs into .claude/skills of the current project.

Files

Subagent-Driven Development

Execute plan by dispatching a fresh implementer subagent per task, a task review (spec compliance + code quality) after each, and a broad whole-branch review at the end.

Why subagents: You delegate tasks to specialized agents with isolated context. By precisely crafting their instructions and context, you ensure they stay focused and succeed at their task. They should never inherit your session's context or history — you construct exactly what they need. This also preserves your own context for coordination work.

Core principle: Fresh subagent per task + task review (spec + quality) + broad final review = high quality, fast iteration

Narration: between tool calls, narrate at most one short line — the ledger and the tool results carry the record.

Continuous execution: Do not pause to check in with your human partner between tasks. Execute all tasks from the plan without stopping. The only reasons to stop are the four named below, or all tasks complete. "Should I continue?" prompts and progress summaries waste their time — they asked you to execute the plan, so execute it.

SKILL.md on GitHub

Details

Repository: github.com/obra/superpowers
Category: AI & Agent Building

More from obra/superpowers

All 14

wind-mcp-skillwind-information-co-ltd/wind-skills用户查询金融数据时触发：A股/港股/美股选股筛选、行情快照、K 线、分钟行情、财务基本面、股东、事件、技术和风险；基金/ETF/LOF 基金筛选、行情、净值、规模、档案、持仓和业绩；指数/板块行情与基本面；债券档案与估值；上市公司公告、财经新闻、宏观经济、汇率和行业指标。不用于台股、日股、韩股、欧股、期货盘口、加密货币104Kskills.sh installs

azure-aimicrosoft/github-copilot-for-azureUse for Azure AI: Search, Speech, OpenAI, Document Intelligence. Helps with search, vector/hybrid search, speech-to-text, text-to-speech, transcription, OCR. WH103.7Kskills.sh installs

microsoft-foundrymicrosoft/github-copilot-for-azureDeploy, evaluate, fine-tune, and manage Foundry agents end-to-end: Docker build, ACR push, hosted/prompt agent create, batch eval, continuous eval, prompt optim103.1Kskills.sh installs

azure-aigatewaymicrosoft/github-copilot-for-azureConfigure Azure API Management as an AI Gateway for AI models, MCP tools, and agents. WHEN: semantic caching, token limit, content safety, load balancing, AI mo103Kskills.sh installs

binance-agentic-walletbinance/binance-skills-hubUse when the user mentions connect/disconnect wallet, sign in, sign out, web3 wallet, wallet address, check balance, how much crypto do I have, send BNB/USDT/cr51.1Kskills.sh installs

replicas-agentreplicas-group/skillBackground coding agent guide for Replicas cloud workspaces with integrations for previews, Slack, Linear, GitHub, Google Workspace, Docker, and media sharing.36.7Kskills.sh installs