kazuki nakai d27c53fa1c
Redesign PM Agent as Self-Improvement Meta-Layer (#421)
* feat: Add PM Agent (Project Manager Agent) for seamless orchestration

Introduces PM Agent as the default orchestration layer that coordinates
all sub-agents and manages workflows automatically.

Key Features:
- Default orchestration: All user interactions handled by PM Agent
- Auto-delegation: Intelligent sub-agent selection based on task analysis
- Docker Gateway integration: Zero-token baseline with dynamic MCP loading
- Self-improvement loop: Automatic documentation of patterns and mistakes
- Optional override: Users can specify sub-agents explicitly if desired

Architecture:
- Agent spec: SuperClaude/Agents/pm-agent.md
- Command: SuperClaude/Commands/pm.md
- Updated docs: README.md (15→16 agents), agents.md (new Orchestration category)

User Experience:
- Default: PM Agent handles everything (seamless, no manual routing)
- Optional: Explicit --agent flag for direct sub-agent access
- Both modes available simultaneously (no user downside)

Implementation Status:
-  Specification complete
-  Documentation complete
-  Prototype implementation needed
-  Docker Gateway integration needed
-  Testing and validation needed

Refs: kazukinakai/docker-mcp-gateway (IRIS MCP Gateway integration)

* feat: Add Agent Orchestration rules for PM Agent default activation

Implements PM Agent as the default orchestration layer in RULES.md.

Key Changes:
- New 'Agent Orchestration' section (CRITICAL priority)
- PM Agent receives ALL user requests by default
- Manual override with @agent-[name] bypasses PM Agent
- Agent Selection Priority clearly defined:
  1. Manual override → Direct routing
  2. Default → PM Agent → Auto-delegation
  3. Delegation based on keywords, file types, complexity, context

User Experience:
- Default: PM Agent handles everything (seamless)
- Override: @agent-[name] for direct specialist access
- Transparent: PM Agent reports delegation decisions

This establishes PM Agent as the orchestration layer while
respecting existing auto-activation patterns and manual overrides.

Next Steps:
- Local testing in agiletec project
- Iteration based on actual behavior
- Documentation updates as needed

* refactor(pm-agent): redesign as self-improvement meta-layer

Problem Resolution:
PM Agent's initial design competed with existing auto-activation for task routing,
creating confusion about orchestration responsibilities and adding unnecessary complexity.

Design Change:
Redefined PM Agent as a meta-layer agent that operates AFTER specialist agents
complete tasks, focusing on:
- Post-implementation documentation and pattern recording
- Immediate mistake analysis with prevention checklists
- Monthly documentation maintenance and noise reduction
- Pattern extraction and knowledge synthesis

Two-Layer Orchestration System:
1. Task Execution Layer: Existing auto-activation handles task routing (unchanged)
2. Self-Improvement Layer: PM Agent meta-layer handles documentation (new)

Files Modified:
- SuperClaude/Agents/pm-agent.md: Complete rewrite with meta-layer design
  - Category: orchestration → meta
  - Triggers: All user interactions → Post-implementation, mistakes, monthly
  - Behavioral Mindset: Continuous learning system
  - Self-Improvement Workflow: BEFORE/DURING/AFTER/MISTAKE RECOVERY/MAINTENANCE

- SuperClaude/Core/RULES.md: Agent Orchestration section updated
  - Split into Task Execution Layer + Self-Improvement Layer
  - Added orchestration flow diagram
  - Clarified PM Agent activates AFTER task completion

- README.md: Updated PM Agent description
  - "orchestrates all interactions" → "ensures continuous learning"

- Docs/User-Guide/agents.md: PM Agent section rewritten
  - Section: Orchestration Agent → Meta-Layer Agent
  - Expertise: Project orchestration → Self-improvement workflow executor
  - Examples: Task coordination → Post-implementation documentation

- PR_DOCUMENTATION.md: Comprehensive PR documentation added
  - Summary, motivation, changes, testing, breaking changes
  - Two-layer orchestration system diagram
  - Verification checklist

Integration Validated:
Tested with agiletec project's self-improvement-workflow.md:
 PM Agent aligns with existing BEFORE/DURING/AFTER/MISTAKE RECOVERY phases
 Complements (not competes with) existing workflow
 agiletec workflow defines WHAT, PM Agent defines WHO executes it

Breaking Changes: None
- Existing auto-activation continues unchanged
- Specialist agents unaffected
- User workflows remain the same
- New capability: Automatic documentation and knowledge maintenance

Value Proposition:
Transforms SuperClaude into a continuously learning system that accumulates
knowledge, prevents recurring mistakes, and maintains fresh documentation
without manual intervention.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: kazuki <kazuki@kazukinoMacBook-Air.local>
Co-authored-by: Claude <noreply@anthropic.com>
2025-10-12 14:22:10 +05:30
2025-09-26 19:37:34 +05:30
2025-09-13 17:28:52 +05:30
2025-08-22 20:39:46 +02:00
2025-09-26 19:37:34 +05:30
2025-09-26 19:37:34 +05:30
2025-08-26 20:30:50 +05:30
2025-09-26 19:37:34 +05:30
2025-09-26 19:37:34 +05:30
2025-09-26 19:37:34 +05:30
2025-09-26 19:37:34 +05:30
2025-10-07 21:57:21 +05:30
2025-09-26 19:37:34 +05:30
2025-09-26 19:37:34 +05:30
2025-09-26 19:37:34 +05:30

🚀 SuperClaude Framework

Transform Claude Code into a Structured Development Platform

Mentioned in Awesome Claude Code Try SuperGemini Framework Try SuperQwen Framework Version License PRs Welcome

Website PyPI npm

English 中文 日本語

Quick StartSupportFeaturesDocsContributing


📊 Framework Statistics

Commands Agents Modes MCP Servers
26 16 7 8
Slash Commands Specialized AI Behavioral Integrations

Use the new /sc:help command to see a full list of all available commands.


🎯 Overview

SuperClaude is a meta-programming configuration framework that transforms Claude Code into a structured development platform through behavioral instruction injection and component orchestration. It provides systematic workflow automation with powerful tools and intelligent agents.

Disclaimer

This project is not affiliated with or endorsed by Anthropic.
Claude Code is a product built and maintained by Anthropic.

Quick Installation

Choose Your Installation Method

Method Command Best For
🐍 pipx pipx install SuperClaude && pipx upgrade SuperClaude && SuperClaude install Recommended - Linux/macOS
📦 pip pip install SuperClaude && pip upgrade SuperClaude && SuperClaude install Traditional Python environments
🌐 npm npm install -g @bifrost_inc/superclaude && superclaude install Cross-platform, Node.js users
⚠️ IMPORTANT: Upgrading from SuperClaude V3

If you have SuperClaude V3 installed, you SHOULD uninstall it before installing V4:

# Uninstall V3 first
Remove all related files and directories :
*.md *.json and commands/

# Then install V4
pipx install SuperClaude && pipx upgrade SuperClaude && SuperClaude install

What gets preserved during upgrade:

  • ✓ Your custom slash commands (outside commands/sc/)
  • ✓ Your custom content in CLAUDE.md
  • ✓ Claude Code's .claude.json, .credentials.json, settings.json and settings.local.json
  • ✓ Any custom agents and files you've added

⚠️ Note: Other SuperClaude-related .json files from V3 may cause conflicts and should be removed.

💡 Troubleshooting PEP 668 Errors
# Option 1: Use pipx (Recommended)
pipx install SuperClaude

# Option 2: User installation
pip install --user SuperClaude

# Option 3: Force installation (use with caution)
pip install --break-system-packages SuperClaude

💖 Support the Project

Hey, let's be real - maintaining SuperClaude takes time and resources.

The Claude Max subscription alone runs $100/month for testing, and that's before counting the hours spent on documentation, bug fixes, and feature development. If you're finding value in SuperClaude for your daily work, consider supporting the project. Even a few dollars helps cover the basics and keeps development active.

Every contributor matters, whether through code, feedback, or support. Thanks for being part of this community! 🙏

Ko-fi

Ko-fi

One-time contributions

🎯 Patreon

Patreon

Monthly support

💜 GitHub

GitHub Sponsors

Flexible tiers

Your Support Enables:

Item Cost/Impact
🔬 Claude Max Testing $100/month for validation & testing
Feature Development New capabilities & improvements
📚 Documentation Comprehensive guides & examples
🤝 Community Support Quick issue responses & help
🔧 MCP Integration Testing new server connections
🌐 Infrastructure Hosting & deployment costs

Note: No pressure though - the framework stays open source regardless. Just knowing people use and appreciate it is motivating. Contributing code, documentation, or spreading the word helps too! 🙏


🎉 What's New in V4

Version 4 brings significant improvements based on community feedback and real-world usage patterns.

🤖 Smarter Agent System

16 specialized agents with domain expertise:

  • PM Agent ensures continuous learning through systematic documentation
  • Deep Research agent for autonomous web research
  • Security engineer catches real vulnerabilities
  • Frontend architect understands UI patterns
  • Automatic coordination based on context
  • Domain-specific expertise on demand

📝 Improved Namespace

/sc: prefix for all commands:

  • No conflicts with custom commands
  • 25 commands covering full lifecycle
  • From brainstorming to deployment
  • Clean, organized command structure

🔧 MCP Server Integration

8 powerful servers working together:

  • Context7 → Up-to-date documentation
  • Sequential → Complex analysis
  • Magic → UI component generation
  • Playwright → Browser testing
  • Morphllm → Bulk transformations
  • Serena → Session persistence
  • Tavily → Web search for deep research
  • Chrome DevTools → Performance analysis

🎯 Behavioral Modes

7 adaptive modes for different contexts:

  • Brainstorming → Asks right questions
  • Business Panel → Multi-expert strategic analysis
  • Deep Research → Autonomous web research
  • Orchestration → Efficient tool coordination
  • Token-Efficiency → 30-50% context savings
  • Task Management → Systematic organization
  • Introspection → Meta-cognitive analysis

Optimized Performance

Smaller framework, bigger projects:

  • Reduced framework footprint
  • More context for your code
  • Longer conversations possible
  • Complex operations enabled

📚 Documentation Overhaul

Complete rewrite for developers:

  • Real examples & use cases
  • Common pitfalls documented
  • Practical workflows included
  • Better navigation structure

🔬 Deep Research Capabilities

Autonomous Web Research Aligned with DR Agent Architecture

SuperClaude v4.2 introduces comprehensive Deep Research capabilities, enabling autonomous, adaptive, and intelligent web research.

🎯 Adaptive Planning

Three intelligent strategies:

  • Planning-Only: Direct execution for clear queries
  • Intent-Planning: Clarification for ambiguous requests
  • Unified: Collaborative plan refinement (default)

🔄 Multi-Hop Reasoning

Up to 5 iterative searches:

  • Entity expansion (Paper → Authors → Works)
  • Concept deepening (Topic → Details → Examples)
  • Temporal progression (Current → Historical)
  • Causal chains (Effect → Cause → Prevention)

📊 Quality Scoring

Confidence-based validation:

  • Source credibility assessment (0.0-1.0)
  • Coverage completeness tracking
  • Synthesis coherence evaluation
  • Minimum threshold: 0.6, Target: 0.8

🧠 Case-Based Learning

Cross-session intelligence:

  • Pattern recognition and reuse
  • Strategy optimization over time
  • Successful query formulations saved
  • Performance improvement tracking

Research Command Usage

# Basic research with automatic depth
/sc:research "latest AI developments 2024"

# Controlled research depth
/sc:research "quantum computing breakthroughs" --depth exhaustive

# Specific strategy selection
/sc:research "market analysis" --strategy planning-only

# Domain-filtered research
/sc:research "React patterns" --domains "reactjs.org,github.com"

Research Depth Levels

Depth Sources Hops Time Best For
Quick 5-10 1 ~2min Quick facts, simple queries
Standard 10-20 3 ~5min General research (default)
Deep 20-40 4 ~8min Comprehensive analysis
Exhaustive 40+ 5 ~10min Academic-level research

Integrated Tool Orchestration

The Deep Research system intelligently coordinates multiple tools:

  • Tavily MCP: Primary web search and discovery
  • Playwright MCP: Complex content extraction
  • Sequential MCP: Multi-step reasoning and synthesis
  • Serena MCP: Memory and learning persistence
  • Context7 MCP: Technical documentation lookup

📚 Documentation

Complete Guide to SuperClaude

🚀 Getting Started 📖 User Guides 🛠️ Developer Resources 📋 Reference
- 📓 [**Examples Cookbook**](Docs/Reference/examples-cookbook.md) *Real-world recipes*

🤝 Contributing

Join the SuperClaude Community

We welcome contributions of all kinds! Here's how you can help:

Priority Area Description
📝 High Documentation Improve guides, add examples, fix typos
🔧 High MCP Integration Add server configs, test integrations
🎯 Medium Workflows Create command patterns & recipes
🧪 Medium Testing Add tests, validate features
🌐 Low i18n Translate docs to other languages

Contributing Guide Contributors


⚖️ License

This project is licensed under the MIT License - see the LICENSE file for details.

MIT License


Star History

Star History Chart

🚀 Built with passion by the SuperClaude community

Made with ❤️ for developers who push boundaries

Back to Top ↑

Description
A configuration framework that enhances Claude Code with specialized commands, cognitive personas, and development methodologies.
Readme MIT 17 MiB
Languages
Python 75.2%
TypeScript 13.7%
Shell 9%
Makefile 2.1%