Second Opinion — Development agent for Claude Code
Gets an independent review from a NON-Claude model (via a local CLI such as kimi, or a local Ollama model) and relays its verdict.
How to install Second Opinion
Installs to ~/.claude/agents/makinggainz-claude-code-token-optimization-second-opinion.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/makinggainz/claude-code-token-optimization/HEAD/agents/second-opinion.md -o ~/.claude/agents/makinggainz-claude-code-token-optimization-second-opinion.md Restart Claude Code, or start a new session, for it to be picked up.
What Second Opinion does
name: second-opinion description: Gets an independent review from a NON-Claude model (via a local CLI such as kimi, or a local Ollama model) and relays its verdict. Use when an independent perspective is worth more than another Claude pass: reviewing a risky change, sanity-checking a design, a second look at a bug that resists diagnosis, or checking whether two different model families agree. Runs on the external model's quota, so it costs almost nothing against the Claude subscription. Read
Alternatives in Development
- Gitnexus Synthesis Critic — GitNexus final review synthesis critic 45.8k ★
- Provider Debugger — Diagnose live-provider compatibility failures for Kimi, GLM, and other OpenAI-compatible endpoints 58 ★
- Lead Triage — Judge ONE confirmation intent packet from lead-desk.py and return exactly three lines - class, would decide, b 45 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
External Reviewer
Gets an independent second opinion on a finished resume or application from a different model (OpenAI Codex CL
Pi
Delegates to the pi CLI in an isolated pi session. Use for work you want done by pi's model and tools rather t
Fleet Plan Alt
Second planning opinion from an independent model. Use when the user asks for another view, when a plan needs
Second Reader
Skeptical structural read of one drafted beat from a prepared brief file. Writes a verdict file, never prose.
Bridge Runner
Cheap wrapper subagent that runs ONE bridge command (from core/bridges/ .md) for a non-Claude model, waits for
Forge Driver
Drives a whole Forge run as the Claude-side driver: grills the owner, gets the plan approved by another model,
Related Skills
Second O Pi Nion
A pi-coding-agent extension that lets the LLM ask a different LLM for a second opinion when it gets stuck.
Visual Second Opinion
Three-pass Gemini visual review — advocate (ship), critic (redesign), synthesis (verdict + cruxes)
Cha Judge
Adjudicate second-eye claims by running them. Every verdict needs a command, an input, and an output — never a