superpowers-mcp
An MCP server that makes superpowers skills available to any LLM that supports the Model Context Protocol.
README
superpowers-mcp
An MCP server that makes superpowers skills available to any LLM that supports the Model Context Protocol.
Superpowers is a skills library for Claude Code that enforces disciplined workflows -- TDD, systematic debugging, brainstorming, planning, and more. This server exposes those skills to any MCP-compatible client: Cursor, Windsurf, Gemini, Kilo Code, Claude Desktop, or your own app.
Quick Start
1. Clone and build
git clone https://github.com/erophames/superpowers-mcp.git
cd superpowers-mcp
npm install
npm run build
2. Run setup
node build/index.js
On first run in a terminal, the setup wizard clones the superpowers repository and saves the configuration.
3. Configure your MCP client
Add to your client's MCP server configuration:
{
"mcpServers": {
"superpowers": {
"command": "node",
"args": ["/absolute/path/to/superpowers-mcp/build/index.js"]
}
}
}
Replace /absolute/path/to/ with the actual path where you cloned the repository.
<details> <summary>Client-specific config file locations</summary>
| Client | Config Location |
|---|---|
| Claude Desktop | ~/Library/Application Support/Claude/claude_desktop_config.json |
| Claude Code | ~/.claude.json or project .mcp.json |
| Cursor | Cursor Settings > MCP |
| Windsurf | ~/.codeium/windsurf/mcp_config.json |
</details>
What It Exposes
The server registers skills through all three MCP primitives:
Tools
| Tool | Description |
|---|---|
list_skills |
List all available skills with descriptions and file lists |
use_skill |
Load a skill by name -- returns the full skill content as instructions to follow (optional guardrail enforcement) |
get_skill_file |
Load a supporting file from a skill (reference docs, prompt templates, scripts) |
recommend_skills |
Recommend top skills for a task using semantic ranking + workflow policy boosts |
compose_workflow |
Build an ordered multi-skill workflow for a goal |
validate_workflow |
Validate selected skills against required workflow guardrails |
semantic_search_skills |
Semantic search across SKILL.md and supporting files |
Prompts
Each skill is registered as an MCP prompt named superpowers:{skill-name}. Clients that support prompt selection will show these in their UI. Selecting a prompt injects the full skill content into the conversation.
Resources
Each skill's SKILL.md and supporting files are registered as resources:
superpowers://skills/brainstorming/SKILL.md
superpowers://skills/test-driven-development/testing-anti-patterns.md
superpowers://skills/systematic-debugging/root-cause-tracing.md
A resource template superpowers://skills/{skillName}/{fileName} enables dynamic access to any file.
Available Skills
| Skill | What it does |
|---|---|
brainstorming |
Collaborative design through questions, approaches, and incremental validation |
writing-plans |
Bite-sized implementation plans with exact file paths and TDD steps |
executing-plans |
Batch execution of plans with review checkpoints between tasks |
subagent-driven-development |
Fresh subagent per task with two-stage code review |
dispatching-parallel-agents |
Distribute independent tasks to concurrent agents |
test-driven-development |
RED-GREEN-REFACTOR cycle -- write failing test first, always |
systematic-debugging |
4-phase root cause analysis: investigate before you fix |
verification-before-completion |
Run the command, read the output, then claim the result |
requesting-code-review |
Dispatch code review and act on severity-categorized feedback |
receiving-code-review |
Technical rigor when receiving feedback -- verify, don't blindly agree |
using-git-worktrees |
Isolated git worktrees for parallel feature development |
finishing-a-development-branch |
Structured options for merge, PR, or cleanup when done |
writing-skills |
Framework for creating, testing, and deploying new skills |
using-superpowers |
How the skill system works and when to invoke skills |
Usage Examples
Once connected, ask your AI assistant:
- "What superpowers skills are available?" (calls
list_skills) - "Use the brainstorming skill to help me design a caching layer."
- "Load the TDD skill and follow it to implement this feature."
- "Read the anti-patterns file from the test-driven-development skill."
- "Recommend the best skills for implementing this MCP feature."
- "Compose a workflow for debugging flaky tests."
- "Validate this workflow: brainstorming, writing-plans, test-driven-development."
- "Search skills for root cause tracing techniques."
Configuration
Environment Variables
| Variable | Description |
|---|---|
SUPERPOWERS_SKILLS_DIR |
Override the skills directory path directly |
Skill Discovery Order
SUPERPOWERS_SKILLS_DIRenvironment variable- Saved directory from the setup wizard (persisted in
~/.config/superpowers-mcp) - Claude Code plugin cache (
~/.claude/plugins/cache/claude-plugins-official/superpowers/*/skills/)
Auto-Updates
When skills are sourced from a git repository, the server checks for updates once per day on startup and pulls new changes automatically.
Development
npm install
npm run build
npm test
| Script | Description |
|---|---|
npm run build |
Compile TypeScript |
npm run dev |
Watch mode compilation |
npm test |
Run all 58 tests |
npm run test:watch |
Watch mode tests |
npm start |
Run the server |
Architecture
src/
index.ts Entry point, stdio transport, setup and update orchestration
server.ts McpServer creation, composes all registrations
config.ts Persistent configuration
update.ts Daily auto-update check
git.ts Git operations (clone, pull, fetch) via execFile
cli/setup.ts Interactive setup wizard
skills/
types.ts Skill, SkillFile, SkillMetadata interfaces
discovery.ts Directory scanning, YAML frontmatter parsing, version resolution
tools/register.ts list_skills, use_skill, get_skill_file
prompts/register.ts One MCP prompt per skill
resources/register.ts Static resources per file + resource template for dynamic access
Tests
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。