SuperClaude Framework

SuperClaude is a configuration framework that layers structured workflows on top of Claude Code. It injects behavioral instructions, slash commands, specialized agents, and optional MCP server integrations to turn Claude Code from a general chat assistant into a more systematic development tool. It's aimed at developers who already use Claude Code daily and want repeatable, documented workflows for research, implementation, testing, and project management instead of ad-hoc prompting.

The project is not affiliated with Anthropic. It's a third-party layer installed via pip/pipx or from source, and it modifies how Claude Code behaves through configuration and command definitions rather than by replacing Claude Code itself.

Key features

• 30 slash commands: cover the development lifecycle, from /sc:brainstorm and /sc:research to /sc:implement, /sc:test, and /sc:pm.
• 20 specialized agents: domain-focused personas (security engineer, frontend architect, PM agent, deep research agent) that Claude Code coordinates automatically based on context.
• 7 behavioral modes: adaptive modes like Brainstorming, Business Panel, Deep Research, Orchestration, Token-Efficiency, Task Management, and Introspection that change how responses are structured.
• 8 MCP server integrations: optional connectors (Tavily, Context7, Sequential-Thinking, Serena, Playwright, Magic, Morphllm-Fast-Apply, Chrome DevTools) for web search, documentation lookup, session memory, and UI generation.
• Deep Research capability: multi-hop web research with adaptive planning strategies (planning-only, intent-planning, unified) and configurable depth from quick to exhaustive.
• CLI tooling: a superclaude command for installing components, managing MCP servers, listing installed pieces, and running diagnostics (superclaude doctor).

Ideal use cases

SuperClaude fits developers running non-trivial projects in Claude Code who want consistent workflows across sessions rather than re-explaining context every time. It's useful for teams that want documented conventions (via PLANNING.md, TASK.md, KNOWLEDGE.md) baked into how Claude Code operates, or for solo developers doing research-heavy work who want a structured /sc:research command instead of manual prompting.

It's a good fit if you already use Claude Code regularly and want more repeatable command-driven behavior, or if you need domain-specific agents (security review, frontend patterns) invoked automatically.

It's not a good fit if you don't use Claude Code at all, since the framework only modifies Claude Code's behavior and has no independent runtime. It's also less useful for quick, one-off tasks where the overhead of installing commands and agents outweighs the benefit, and the optional MCP servers add extra setup if you want the performance gains they offer.

Installation

The recommended path uses pipx:

# Install from PyPI
pipx install superclaude

# Install commands (installs all 30 slash commands)
superclaude install

# Install MCP servers (optional, for enhanced capabilities)
superclaude mcp --list
superclaude mcp
superclaude mcp --servers tavily --servers context7

# Verify installation
superclaude install --list
superclaude doctor

Alternatively, install directly from source:

git clone https://github.com/SuperClaude-Org/SuperClaude_Framework.git
cd SuperClaude_Framework
./install.sh

After installing, restart Claude Code to load the 30 new commands. MCP servers are optional but claimed to give 2-3x faster execution and 30-50% fewer tokens on supported workflows; the framework works fully without them, just at standard performance. A TypeScript plugin system is planned for a future major version but isn't available yet, so installation currently relies on the Python CLI and slash commands described above.