Claude Cognitive Reasoning Framework

A comprehensive cognitive architecture for AI agents, featuring 9 reasoning patterns, a meta-orchestrator, and production-ready agent implementations. This framework moves beyond "prompt engineering" into "cognitive architecture" — ensuring agents don't use a hammer for a screw.

Author: Kim Asplund License: MIT Version: 3.0

---

Features

• 9 Core Reasoning Patterns - Specialized methodologies for different problem types
• IR-v2 Meta-Orchestrator - Automatic pattern selection using 11-dimension scoring
• 17 Production Agents - Ready-to-use agent implementations with ChromaDB memory
• 10 Skill Frameworks - Domain-specific capabilities (security, git, documentation, etc.)
• Parallel Execution - Run multiple patterns simultaneously for 2-4x speedup
• Handover Protocol - State persistence across multi-pattern reasoning chains
• Self-Improving Architecture - Agents learn and improve via ChromaDB memory
• Memory Consolidation - "Sleep-like" periodic consolidation for cross-agent learning
• Autonomous Orchestration - Auto-invoke skills/agents based on task patterns with hooks

---

Quick Start

One-Line Installation

git clone https://github.com/kimasplund/claude_cognitive_reasoning.git && cd claude_cognitive_reasoning && ./scripts/install.sh

Manual Installation

Clone the repository and copy files to your Claude Code configuration directory:

# Clone the repository
git clone https://github.com/kimasplund/claude_cognitive_reasoning.git
cd claude_cognitive_reasoning

# Or use the install script
./scripts/install.sh          # Full installation
./scripts/install.sh --skills # Skills only
./scripts/install.sh --agents # Agents only

# Manual: Install cognitive skills (personal - available in all projects)
cp -r cognitive-skills/* ~/.claude/skills/

# Install agents
cp agent-examples/*.md ~/.claude/agents/

# Install skill frameworks
cp -r skill-frameworks/* ~/.claude/skills/

# Install commands
mkdir -p ~/.claude/commands
cp commands/*.md ~/.claude/commands/

# Copy the CLAUDE.md quick reference (optional - for project-level)
cp CLAUDE.md ~/.claude/CLAUDE.md

Verify Installation

# Check skills are installed
ls ~/.claude/skills/

# You should see directories like:
# integrated-reasoning-v2/
# tree-of-thoughts/
# adversarial-reasoning/
# ...

# Check agents are installed
ls ~/.claude/agents/

# You should see files like:
# ceo-orchestrator.md
# architect-agent.md
# security-agent.md
# ...

Start Using

Skills activate automatically when Claude detects a matching problem type. You can also invoke them manually:

# In Claude Code CLI, skills are available as slash commands
/integrated-reasoning-v2    # Meta-orchestrator for complex problems
/tree-of-thoughts           # Systematic solution exploration
/adversarial-reasoning      # Security validation and stress-testing

# Agents are available via the Task tool
# Claude will automatically use appropriate agents based on context

---

Directory Structure

claude_cognitive_reasoning/
├── CLAUDE.md                          # Quick reference guide
├── README.md                          # This file
│
├── cognitive-skills/                  # Core reasoning patterns
│   ├── INTEGRATION_GUIDE.md           # How to compose patterns
│   ├── integrated-reasoning-v2/       # Meta-orchestrator (IR-v2)
│   ├── tree-of-thoughts/              # ToT - Optimization
│   ├── breadth-of-thought/            # BoT - Exploration
│   ├── self-reflecting-chain/         # SRC - Sequential logic
│   ├── hypothesis-elimination/        # HE - Diagnosis
│   ├── adversarial-reasoning/         # AR - Security validation
│   ├── dialectical-reasoning/         # DR - Trade-offs
│   ├── analogical-transfer/           # AT - Novel problems
│   ├── rapid-triage-reasoning/        # RTR - Time-critical
│   ├── negotiated-decision-framework/ # NDF - Stakeholder alignment
│   ├── parallel-execution/            # Run patterns simultaneously
│   ├── reasoning-handover-protocol/   # Cross-pattern state transfer
│   └── ralph-loop-integration/        # Iterative refinement wrapper
│
├── agent-examples/                    # Production agent implementations
│   ├── ceo-orchestrator.md            # Strategic decision-making
│   ├── architect-agent.md             # System design (DR, ToT, AR)
│   ├── security-agent.md              # Threat modeling (AR, HE)
│   ├── root-cause-analyzer.md         # Debugging (HE, ChromaDB)
│   ├── break-it-tester.md             # Adversarial QA (AR)
│   ├── qa-agent.md                    # Test planning
│   ├── pm-planner.md                  # Project planning
│   ├── pm-executor.md                 # Execution tracking
│   ├── research-specialist.md         # Research synthesis
│   ├── agent-hr-manager.md            # Agent creation & management
│   ├── memory-consolidation-agent.md  # "Sleep-like" memory optimization
│   

...