Codex Reasoning Skill

Purpose

Leverage GPT-5-Codex's different reasoning patterns to get alternative perspectives, second opinions, and specialized algorithmic solutions.

Unique Capability

What This Adds: Different AI reasoning:

• GPT-5-Codex optimized for agentic coding
• Different algorithmic approaches
• Alternative architecture perspectives
• Fast one-shot reasoning

When to Use

Perfect For:

• Getting second opinion on architecture
• Exploring alternative implementations
• Algorithmic optimization problems
• When stuck (different perspective helps)
• Comparing solution approaches
• Performance-critical algorithms

Don't Use When:

• Claude's solution is clearly working
• Simple tasks (no need for alternatives)
• Consistency with existing Claude code matters

Usage

# Second opinion
/codex-reasoning "What's an alternative approach to user authentication?"

# Algorithm comparison
/codex-reasoning "Optimize this sorting algorithm for large datasets"

# Architecture alternative
/codex-reasoning "What's an alternative to microservices for our scale?"

CLI Command

codex
> /model gpt-5-codex
> "Your reasoning question"

# Or directly
codex "What's an alternative approach to X?"

Why Use Both Models?

Real Example

Architecture Decision:
- Claude suggests: Event-driven with message queue
- GPT-5-Codex suggests: REST with polling + webhooks

Result: Hybrid approach combining benefits of both

Integration Pattern

// Get both perspectives
const claudeApproach = await claudeArchitect("Design auth system");
const codexApproach = await codexReasoning("Alternative auth approach?");

// Use LLM Council for consensus
const consensus = await llmCouncil(`
  Claude: ${claudeApproach}
  Codex: ${codexApproach}
  Which is better for our use case?
`);

Memory Integration

• Key: multi-model/codex/reasoning/{task_id}
• Tags: WHO=codex-cli:reasoning, WHY=alternative-perspective