Gene: Reasoning Chain

Description

Advanced reasoning skill implementing chain-of-thought and tree-of-thought reasoning patterns. Enhanced with best practices from Anthropic's skills-main for complex problem solving.

Trigger Conditions

• User asks to "think step by step" or "reason through"
• Complex multi-step problems requiring analysis
• Mathematical or logical reasoning tasks
• Decision making with multiple factors
• Problem decomposition requests
• "analyze", "evaluate", "reason through"

Reasoning Patterns

Chain of Thought (CoT)

For sequential reasoning problems:

1. Deconstruct - Break problem into atomic steps
2. Validate - Check each step before proceeding
3. Synthesize - Combine steps into final answer
4. Verify - Cross-check result against original problem

Tree of Thoughts (ToT)

For branching decision problems:

1. Explore - Generate multiple solution paths
2. Evaluate - Assess each branch's viability
3. Prune - Eliminate weak branches
4. Backtrack - Revert if path fails
5. Commit - Select optimal path

Execution Protocol

Step 1: Problem Analysis

• Identify problem type (deductive, inductive, abductive)
• List known constraints and variables
• Determine required reasoning depth
• Select reasoning pattern (CoT vs ToT)

Step 2: Step Execution

• Execute one reasoning step at a time
• Validate each step before advancing
• Track intermediate conclusions
• Flag assumptions explicitly

Step 3: Synthesis

• Combine steps into coherent reasoning chain
• Identify logical gaps
• Verify conclusion follows from premises
• Prepare structured output

Output Format

Present reasoning in clear, numbered steps. Use:

• Bold for key insights
• Tables for comparative analysis
• Code blocks for mathematical notation
• Blockquotes for assumptions

Guardrails

• Never skip validation steps
• Explicitly state assumptions
• Acknowledge uncertainty
• Flag when additional context needed

Integration

• Works with all other genes
• Priority: High when complex reasoning required