Gold Reviewer — Data & AI agent for Claude Code
Scores report-engine changes against the gold sets (evals/gold/) by replaying exported fixtures on a scratch database, and reports recall and contradictions per fact.
How to install Gold Reviewer
Installs to ~/.claude/agents/aman4563-finresearch-gold-reviewer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Aman4563/finresearch/HEAD/.claude/agents/gold-reviewer.md -o ~/.claude/agents/aman4563-finresearch-gold-reviewer.md Restart Claude Code, or start a new session, for it to be picked up.
What Gold Reviewer does
name: gold-reviewer description: Scores report-engine changes against the gold sets (evals/gold/) by replaying exported fixtures on a scratch database, and reports recall and contradictions per fact. Never edits gold facts to pass. tools: Read, Grep, Glob, Bash model: inherit
You measure whether a report-engine change (prompts, roles, verification, the publish gate, renderers, `evals/`) keeps the gold-set bar: at least 90% key-fact recall and 0 contradicted facts.
Rules:
- No live rese
Alternatives in Data & AI
- Patina Detector — Triggers when Claude needs to find AI-writing patterns or suspect zones in KO/EN/ZH/JA text 352 ★
- News Sentiment Analyst — Specialized agent for news sentiment analysis of B3 stocks 124 ★
- Book Evaluator — Independent evaluator for the book pipeline 114 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Voice Implementer
Second-pass writer. Takes the draft-implementer's output and revises for voice consistency, rhythm precision,
Flashcard Writer
Writes or refreshes Trial Recall flashcards from the Notion oncology databases. Use after a trial is added or
Context Quality Reviewer
Reads CLAUDE.md and .claude/rules/ for a project and evaluates quality via LLM-judgment: length, internal cont
Prompter
Use this agent when you need to create, refine, or optimize prompts for AI systems. This includes writing new
Alphalattice Data
Advise on one product-exported Data/Feature issue or update state, preserving its evidence and permission limi
Conversation Assessor
Reads one exported batch of conversations and writes one verdict file inside work/results/. Use it only for tr
Related Skills
Write Skill
Author a new skill from scratch with paired trigger fixtures and full validation. Use when adding a skill that
Companion Vault
An Obsidian vault and setup kit that gives your AI assistant a lasting memory. It learns how you work and writ
Brain Reconcile
Find and resolve contradictions across the second brain, keeping a history trail. Never edits preserved or raw