secondsky__claude-skills/plugins/multi-ai-consultant

Multi-AI Consultant

Get second opinions from Gemini, OpenAI, or fresh Claude when stuck or making critical decisions.

---

Quick Start

# Install skill
cd ~/.claude/skills
git clone [this-repo] multi-ai-consultant

# Setup APIs
export GEMINI_API_KEY="your-key"  # Required
export OPENAI_API_KEY="sk-..."    # Optional

# Copy templates to your project
cp ~/.claude/skills/multi-ai-consultant/templates/GEMINI.md ./
cp ~/.claude/skills/multi-ai-consultant/templates/codex.md ./
cp ~/.claude/skills/multi-ai-consultant/templates/.geminiignore ./

# Start using
# Claude Code will automatically suggest consultations when stuck

---

Auto-Trigger Keywords

This skill activates when you say:

Stuck/Need Help

• "I'm stuck"
• "still not working"
• "this isn't working"
• "can't figure out"
• "no idea why"
• "need help"
• "need a second opinion"
• "want another perspective"

Architecture/Design

• "architecture decision"
• "how should we structure"
• "which approach is better"
• "design question"
• "choosing between"
• "what's the best way to"

Security

• "is this secure"
• "security concern"
• "vulnerability"
• "authentication"
• "authorization"
• "validation"

Manual Commands

• /consult-gemini [question]
• /consult-codex [question]
• /consult-claude [question]
• /consult-ai [question]

---

The Three AIs

AI	Best For	Cost	Speed
Gemini 2.5 Pro	Web research, latest docs, thinking	~$0.10-0.50	Slower
OpenAI Codex	Repo-aware analysis, code review	~$0.05-0.30	Medium
Fresh Claude	Quick opinion, fresh perspective	Free	Fast

---

How It Works

1. You get stuck on a bug after one attempt
2. Claude suggests consulting another AI
3. You approve
4. External AI analyzes the problem with fresh eyes
5. Claude synthesizes both perspectives in 5-part format:
   • 🤖 Claude's analysis
   • 💎/🔷/🔄 Other AI's analysis
   • 🔍 Key differences
   • ⚡ Synthesis
   • ✅ Recommended action
6. You decide which approach to take

Result: Better solutions, faster debugging, fresh perspectives

---

Example

You: "Fix this JWT auth bug"

Claude: Tries token expiry fix → Still failing

Claude: "I've tried one approach without success. Should I consult Gemini for a fresh perspective?"

You: "yes"

Claude: Executes /consult-gemini internally

Gemini: Searches web → Finds Cloudflare Workers don't support process.env → Suggests env.JWT_SECRET

Claude: Synthesizes both analyses → Shows 5-part comparison → Asks permission

You: "proceed"

Claude: Implements fix → ✅ Working

Time saved: 20-30 minutes of trial-and-error

---

Setup

1. Install CLIs

# Gemini (required)
npm install -g @google/generative-ai-cli
export GEMINI_API_KEY="your-key"
# Get key: https://aistudio.google.com/apikey

# Codex (optional)
npm install -g codex
export OPENAI_API_KEY="sk-..."
# Get key: https://platform.openai.com/api-keys

2. Install Skill

cd ~/.claude/skills
git clone [this-repo] multi-ai-consultant

3. Copy Templates

# In your project root
cp ~/.claude/skills/multi-ai-consultant/templates/GEMINI.md ./
cp ~/.claude/skills/multi-ai-consultant/templates/codex.md ./
cp ~/.claude/skills/multi-ai-consultant/templates/.geminiignore ./

4. Test

# Test Gemini
gemini -p "test" && echo "✅ Working"

# Test Codex (if installed)
codex exec "test" --yolo && echo "✅ Working"

# Ask Claude Code
# "I'm stuck on this bug"
# → Should suggest consultation

---

Usage

Automatic (Recommended)

Just work normally. Claude Code will suggest consultations when:

• Stuck after one failed attempt
• Making architectural decisions
• Security concerns

Manual

# Specific AI
/consult-gemini Is this JWT implementation secure?
/consult-codex Review this codebase for code smells
/consult-claude Quick sanity check on this approach

# Let Claude choose
/consult-ai How should we structure this microservices architecture?

---

Features

✅ Automatic triggers - Suggests consultation when stuck ✅ Forced synthesis - Prevents parroting external AI ✅ Smart context - Selective file inclusion ✅ Cost tracking - Logs every consultation ✅ Privacy protected - Respects .gitignore + .geminiignore ✅ Three AI options - Gemini (web search), Codex (repo-aware), Fresh Claude (free) ✅ Pre-flight checks - Verifies CLIs work before calling ✅ Error handling - Helpful messages for common issues

---

Cost

Typical consultation:

• Gemini: ~$0.10-0.50 (depending on context size)
• Codex: ~$0.05-0.30 (depending on codebase size)
• Fresh Claude: Free

View spending:

consultation-log-parser.sh --summary

Estimated ROI: 20-30 minutes saved per bug = worth the cost

---

Privacy

Automatic protection:

• Both CLIs respect .gitignore
• Never sends node_modules/, .env*, etc.

Extra protection:

• Create .geminiignore for additional exclusions
• Claude warns if sensitive files detected

You control:

• Approve before each consultation
• See what context will be sent
• Choose which AI (or decline)

---

Why This Works

Problem: Claude Code sometimes gets stuck in a mental rut or lacks latest info

Solution: External AIs provide:

• Fresh perspective (different reasoning)
• Web research (Gemini's Google Search)
• Repo-aware analysis (Codex's auto-scan)
• Latest documentation (Gemini's grounding)

Result: Better solutions, faster debugging, informed decisions

---

When NOT to Use

❌ Simple bugs you haven't tried fixing yet ❌ Typos or syntax errors ❌ Questions answered by reading docs ❌ Budget is absolute $0 (use Fresh Claude only)

---

Troubleshooting

Gemini not working?

npm install -g @google/generative-ai-cli
export GEMINI_API_KEY="your-key"

Codex not working?

npm install -g codex
export OPENAI_API_KEY="sk-..."

Skill not discovered?

ls -la ~/.claude/skills/multi-ai-consultant
# Check SKILL.md exists and has YAML frontmatter

Too expensive?

• Use Fresh Claude (free) for quick opinions
• Use Gemini only for complex problems
• Check logs: consultation-log-parser.sh --summary

---

Documentation

• Full docs: See SKILL.md
• Planning: See planning/multi-ai-consultant-*.md
• CLI reference: See references/
• Examples: See examples/

---

Contributing

Found an issue? Document in SKILL.md Known Issues

Want to add an AI? Create commands/consult-newai.md

Improving synthesis? Edit templates/GEMINI.md or templates/codex.md

---

License

MIT

---

Last Updated: 2025-11-07 Status: Production Ready Maintainer: Claude Skills Maintainers | maintainers@example.com