Multi-Model Second Opinion Council
Get blind second opinions from Gemini, GPT, or any OpenRouter model, then synthesize one answer that states what changed and why.
Before & after
Same prompt, same agent — the only difference is the skill.
I can review your code and give you feedback. If you want a second opinion, I could think about it from a different angle and point out potential issues. You might also want to ask other developers for their perspective, which often helps catch things that one person misses. Getting multiple viewpoints on important decisions is always a good idea. Would you like me to review it, or would you prefer to gather feedback from multiple people? I can give you my thoughts and maybe suggest what questions to ask others.
MULTI-MODEL AUDIT REPORT CODE SUBMISSION • Function: pagination handler (sort + cursor logic) • Context: handles 10k+ row datasets; performance-sensitive • Claude assessment: "Clean iteration; edge case on cursor null" MODEL REVIEW SUMMARY CLAUDE (Primary model) • Rating: 7/10 | Concern: cursor null handling incomplete • Feedback: Add guard clause for null cursors; current code crashes on first page • Suggestion: Test with empty result set before deploy GEMINI (Secondary model via OpenRouter) • Rating: 8/10 | Concern: performance at scale • Feedback: Index on sort_key column critical; query plan will fail without it • Suggestion: Add database query explain comment; include migration script CODEX (Third model) • Rating: 6/10 | Concern: API contract not documented • Feedback: Cursor format not declared; callers guessing format • Suggestion: Add JSDoc block with cursor format spec (base64 encoded timestamp) SYNTHESIS (Consensus) ✗ HIGH PRIORITY: Add cursor null guard + empty result test (Claude + Codex agree) ✗ HIGH PRIORITY: Add database index; include in migration (Gemini warning) ✓ NICE TO HAVE: Add JSDoc documentation (Codex suggestion) FINAL RECOMMENDATION • Merge after: null guard + migration script added • Estimated fix time: 1 hour • Confidence in revised code: 9/10 (all 3 models agree on fixes)
About this skill
Claude stays the one doing the work while other models review it independently. You get a real second opinion, not an echo, and nothing sensitive leaves the machine by accident.
What it does
- Checks access and redacts secrets and personal data before any outbound call
- Asks each model blind first, then runs a targeted critique only where views differ
- Weighs disagreements on evidence and verifies checkable claims before adopting them
- Calls models through OpenRouter or installed vendor CLIs and saves raw responses for audit
What you get
A second-opinion report: bottom line, agreements, a disagreement table with evidence, adopted and rejected changes, what data was sent, and cost when known.
Best for
Developers and operators making high-stakes or irreversible calls (migrations, architecture, security) who want a cross-check from another vendor's model.
SKILL.md preview
---
name: council-skill
description: Get an independent second opinion from other models (Gemini, GPT/Codex, or any model on OpenRouter) on a plan, diff, or answer, while Claude stays the one doing the work - package a minimal redacted brief, ask each model blind before showing Claude's view, compare where they agree and disagree, and synthesize one answer that says exactly what changed. Use when the user says "get a second opinion from Gemini", "ask GPT what it thinks", "run this past another model", "council review this migration plan", "have Codex check this diff", or casually "am I missing something here? check with another model".
---
# Multi-Model Second Opinion Council
… (sign up to view the full skill)More ai & agents skills
View all AI & Agents skills →New Skill Scaffolder
Create a new Claude Code skill correctly: right mechanism, valid name and folder, a triggering description, and a smoke test.
Skill Library Lint and Overlap Report
Lint every skill in a repo at once: frontmatter, name-folder match, weak descriptions, overlapping triggers, broken refs, plus a CI script.
Self-Verifying Agent Loop: Plan, Check, Retry, Close
Turns a goal into a task plan with machine-run checks, executes with retry caps, escalates, and refuses to close until verified.
Skill Test Runner (Trigger and A/B)
Test a skill in fresh sessions: ready-to-run prompts, commands, and rubric for trigger recall, false triggers, and blind A/B.