second-opinion

name	second-opinion
description	Get second opinion from OpenAI (Codex), Google (Gemini), or Anthropic (Claude Code) models. Use when validating architectural decisions, reviewing code security, comparing implementation approaches, or when multi-model consensus adds value. Supports all three major AI CLI tools for maximum flexibility.

Second Opinion via Codex, Gemini & Claude Code CLI

Get external AI perspective from OpenAI (via Codex CLI), Google (via Gemini CLI), or Anthropic (via Claude Code CLI) to validate decisions or compare approaches.

Quick Patterns

Using Codex (OpenAI)

Simple Question

codex exec -m gpt-5.3-codex --output-last-message /tmp/claude/answer.txt "Your question here"
cat /tmp/claude/answer.txt

Structured Analysis (Recommended)

IMPORTANT: When using --output-schema, ALL objects (including nested ones) must have:

• "additionalProperties": false
• "required": [...] with all property names

cat > /tmp/claude/schema.json << 'EOF'
{
  "type": "object",
  "properties": {
    "assessment": { "type": "string" },
    "strengths": { "type": "array", "items": { "type": "string" } },
    "concerns": { "type": "array", "items": { "type": "string" } },
    "recommendation": { "type": "string" }
  },
  "required": ["assessment", "strengths", "concerns", "recommendation"],
  "additionalProperties": false
}
EOF

codex exec -m gpt-5.3-codex --output-schema /tmp/claude/schema.json \
  --output-last-message /tmp/claude/result.json \
  "Analyze [topic]. Provide structured assessment."

cat /tmp/claude/result.json

Nested Objects Example

# Nested objects MUST also have additionalProperties: false
cat > /tmp/claude/nested_schema.json << 'EOF'
{
  "type": "object",
  "properties": {
    "summary": { "type": "string" },
    "details": {
      "type": "object",
      "properties": {
        "score": { "type": "string" },
        "items": { "type": "array", "items": { "type": "string" } }
      },
      "required": ["score", "items"],
      "additionalProperties": false
    }
  },
  "required": ["summary", "details"],
  "additionalProperties": false
}
EOF

Using Gemini (Google)

Simple Question

gemini -p "Your question here" --output-format text > /tmp/claude/answer.txt
cat /tmp/claude/answer.txt

JSON Output

gemini -p "Analyze [topic]. Respond in JSON with: assessment (string), strengths (array), concerns (array), recommendation (string)" \
  --output-format json > /tmp/claude/result.json
cat /tmp/claude/result.json

Note: Gemini CLI doesn't support JSON Schema validation like Codex. For strict structured output, use Codex with --output-schema.

Using Claude Code (Anthropic)

Simple Question

claude -p "Your question here" --model claude-opus-4-6 --output-format text > /tmp/claude/answer.txt
cat /tmp/claude/answer.txt

JSON Output

claude -p "Analyze [topic]. Respond in JSON with: assessment (string), strengths (array), concerns (array), recommendation (string)" \
  --model claude-opus-4-6 --output-format json > /tmp/claude/result.json
cat /tmp/claude/result.json

With Cheaper Model (Simple Tasks)

claude -p "Your question here" --model claude-sonnet-4-5-20250929 --output-format text > /tmp/claude/answer.txt
cat /tmp/claude/answer.txt

Note: Claude Code CLI uses --print (-p) for non-interactive mode. The spawned instance runs independently without access to the current session context, providing a genuinely fresh perspective. JSON output via --output-format json wraps the response in a structured message object.

Use Cases

Architecture Review

Codex:

codex exec -m gpt-5.3-codex --output-last-message /tmp/claude/arch.txt \
  "Review this architecture decision: [description].
   Assess: scalability, maintainability, security risks, alternatives."
cat /tmp/claude/arch.txt

Gemini:

gemini -p "Review this architecture decision: [description].
  Assess: scalability, maintainability, security risks, alternatives." \
  --output-format text > /tmp/claude/arch.txt
cat /tmp/claude/arch.txt

Claude Code:

claude -p "Review this architecture decision: [description].
  Assess: scalability, maintainability, security risks, alternatives." \
  --model claude-opus-4-6 --output-format text > /tmp/claude/arch.txt
cat /tmp/claude/arch.txt

Security Audit

Codex:

codex exec -m gpt-5.3-codex --output-last-message /tmp/claude/security.txt \
  "Security review of [file/code]:
   - Input validation
   - Authentication/authorization
   - Data exposure risks
   Provide specific vulnerabilities and fixes."
cat /tmp/claude/security.txt

Gemini:

gemini -p "Security review of [file/code]:
  - Input validation
  - Authentication/authorization
  - Data exposure risks
  Provide specific vulnerabilities and fixes." \
  --output-format text > /tmp/claude/security.txt
cat /tmp/claude/security.txt

Claude Code:

claude -p "Security review of [file/code]:
  - Input validation
  - Authentication/authorization
  - Data exposure risks
  Provide specific vulnerabilities and fixes." \
  --model claude-opus-4-6 --output-format text > /tmp/claude/security.txt
cat /tmp/claude/security.txt

Code Review

Codex:

codex exec -m gpt-5.3-codex --output-last-message /tmp/claude/review.txt \
  "Review [file] for: bugs, performance issues, maintainability.
   Provide line-level recommendations."
cat /tmp/claude/review.txt

Gemini:

gemini -p "Review [file] for: bugs, performance issues, maintainability.
  Provide line-level recommendations." \
  --output-format text > /tmp/claude/review.txt
cat /tmp/claude/review.txt

Claude Code:

claude -p "Review [file] for: bugs, performance issues, maintainability.
  Provide line-level recommendations." \
  --model claude-opus-4-6 --output-format text > /tmp/claude/review.txt
cat /tmp/claude/review.txt

Key Options

Codex CLI Options

Option	Purpose
-m gpt-5.3-codex	Best model (recommended)
-m o4-mini	Faster, cheaper for simple tasks
--output-schema file.json	Structured JSON with schema validation
--output-last-message file.txt	Save response to file
-i image.png	Include image for analysis

Gemini CLI Options

Option	Purpose
--model gemini-3-pro-preview	Best reasoning (explicit selection)
--model gemini-3-flash-preview	Fast and capable
Auto-routing (default)	CLI selects best model automatically
--output-format text	Plain text output
--output-format json	JSON output (no schema validation)
-p "prompt"	Non-interactive mode (required)

Claude Code CLI Options

Option	Purpose
--model claude-opus-4-6	Most capable model (recommended)
--model claude-sonnet-4-5-20250929	Balanced speed and quality
--model claude-haiku-4-5-20251001	Fastest, cheapest for simple tasks
--output-format text	Plain text output (recommended)
--output-format json	JSON message object output
--output-format stream-json	Streaming JSON output
-p "prompt"	Non-interactive print mode (required)
--max-turns N	Limit agentic turns (default: no limit)

Provider Comparison

Use Case	Codex (OpenAI)	Gemini (Google)	Claude Code (Anthropic)
Complex architectural review	gpt-5.3-codex	gemini-3-pro-preview or auto	claude-opus-4-6
Fast code review	o4-mini	gemini-3-flash-preview or auto	claude-haiku-4-5-20251001
Security audit (deep)	gpt-5.3-codex	Auto-routing (recommended)	claude-opus-4-6
Structured output (schema)	gpt-5.3-codex + schema	N/A (no schema support)	N/A (no schema support)
Balanced quality/speed	gpt-5.3-codex	Auto-routing	claude-opus-4-6
Multi-provider consensus	All three! Run all and compare	All three! Run all and compare	All three! Run all and compare

Presenting Results

1. Label clearly which provider was used:
   • "Second opinion (OpenAI/Codex - gpt-5.3-codex)"
   • "Second opinion (Google/Gemini - gemini-3-pro)"
   • "Second opinion (Anthropic/Claude Code - claude-opus-4-6)"
   • "Consensus (Codex + Gemini + Claude Code)" if using all three
2. Compare with your own analysis
3. Highlight areas of agreement and disagreement
4. Synthesize recommendation based on multiple perspectives
5. Optional: Get consensus by asking all providers and comparing outputs

Prerequisites

Verify CLIs are available before use:

# Check Codex
codex --version || echo "Codex CLI not installed"

# Check Gemini
gemini --version || echo "Gemini CLI not installed"

# Check Claude Code
claude --version || echo "Claude Code CLI not installed"

Authentication:

• Codex: codex login or set OPENAI_API_KEY env var
• Gemini: Already authenticated via Google OAuth or set GEMINI_API_KEY env var
• Claude Code: Already authenticated via claude login or set ANTHROPIC_API_KEY env var

Multi-Provider Consensus Example

Get second opinions from all three providers and compare:

# 1. Ask Codex
codex exec -m gpt-5.3-codex --output-last-message /tmp/claude/codex_opinion.txt \
  "Should we use Redis or PostgreSQL for session storage in e-commerce app?"

# 2. Ask Gemini
gemini -p "Should we use Redis or PostgreSQL for session storage in e-commerce app?" \
  --output-format text > /tmp/claude/gemini_opinion.txt

# 3. Ask Claude Code
claude -p "Should we use Redis or PostgreSQL for session storage in e-commerce app?" \
  --model claude-opus-4-6 --output-format text > /tmp/claude/claude_opinion.txt

# 4. Compare outputs
echo "=== Codex (OpenAI) ===" && cat /tmp/claude/codex_opinion.txt
echo ""
echo "=== Gemini (Google) ===" && cat /tmp/claude/gemini_opinion.txt
echo ""
echo "=== Claude Code (Anthropic) ===" && cat /tmp/claude/claude_opinion.txt

Analyze agreement/disagreement across all three providers and synthesize final recommendation.