Cold Diff Reviewer — Development agent for Claude Code
Independent, framing-free review of a diff.
How to install Cold Diff Reviewer
Installs to ~/.claude/agents/raaif-yousuf-agent-orchestration-cold-diff-reviewer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Raaif-Yousuf/agent-orchestration/HEAD/agents/cold-diff-reviewer.md -o ~/.claude/agents/raaif-yousuf-agent-orchestration-cold-diff-reviewer.md Restart Claude Code, or start a new session, for it to be picked up.
What Cold Diff Reviewer does
name: cold-diff-reviewer description: Independent, framing-free review of a diff. Use when a change needs a second opinion that was not primed by the author's reasoning, the ticket text, or the orchestrator's own theory of the fix. tools: Read, Grep, Glob model: sonnet
You are reviewing a diff. You have been given ONLY the diff and the file paths it touches: no ticket, no author's reasoning, no orchestrator theory about what the change fixes or why. That omission is deliberate and load-
Alternatives in Development
- Finding Checker — Blind adversarial checker — given ONLY a finding artifact and its evidence (never the author's reasoning), tri 348 ★
- Guidelines Extractor — Given a URL or raw text of journal author guidelines, extracts structured requirements into a guidelines/jour 67 ★
- 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
Fleet Plan Alt
Second planning opinion from an independent model. Use when the user asks for another view, when a plan needs
Pies Evaluator
Cold, read-only PIES second opinion for escalation-triggered nodes; reruns verification against exact criteria
Executor Opus Medium
Ticket executor — Opus at medium reasoning effort, for a fully specified mechanical ticket that still wants Op
Codex Consultant
An independent code review and debugging subagent that uses Codex MCP as its sole execution engine, acting as
Positioning Critic
Adversarial second opinion on an overnight positioning delta (node P, autonomous mode). Independent context fr
Second Opinion
Gets an independent review from a NON-Claude model (via a local CLI such as kimi, or a local Ollama model) and
Related Skills
Codex Consult
Consult OpenAI Codex (codex exec) for a second opinion on a prompt. Handles the argv-hang, stdin-mode, reasoni
FileMaker Second Opinion
A reasoning skill for AI-assisted FileMaker development, built from production experience. No techniques, no k
Agy Review
Adversarial critique via agy using Gemini Pro — independent second opinion on code, plans, or arguments