Loyal Evaluator — Testing & QA agent for Claude Code
Spec-starved reconstruction of what a product actually does and what it seems to be for, run by /gate and /loyal through run_isolated.py inside a blind copy with every intent file removed.
How to install Loyal Evaluator
Installs to ~/.claude/agents/mae7777-coding-pipeline-loyal-evaluator.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/mAE7777/coding-pipeline/HEAD/agents/loyal-evaluator.md -o ~/.claude/agents/mae7777-coding-pipeline-loyal-evaluator.md Restart Claude Code, or start a new session, for it to be picked up.
What Loyal Evaluator does
name: loyal-evaluator description: Spec-starved reconstruction of what a product actually does and what it seems to be for, run by /gate and /loyal through run_isolated.py inside a blind copy with every intent file removed. Pass 1 receives only who the user is and how to run the product; pass 2 receives the one-line goal. Grounds every behavior in a real trace. Not for direct use. tools: Read, Grep, Glob, Bash model: inherit effort: xhigh
You have been handed an unfamiliar product and a
Alternatives in Testing & QA
- Evaluator Cosmwasm Agent — CosmWasm smart-contract vulnerability evaluator — spawned per smart_contract surface with chain_family=cosmwas 97 ★
- Forge Speccer Validator — Pre-planning path-validation gate 55 ★
- Uxaudit Scout — Project Scout for the uxaudit pipeline 52 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Live Evaluator
Phase 3 of the grounded-persona-eval method — drives one persona through a real, live interaction with the pro
QA Author
Blind QA-plan author. Spawned by /blueprint immediately after context.md is written — reads ONLY context.md an
Qc Lead
Use this agent when the qc-engineer has finished a test pass and produced an evidence set, when a run needs it
Worker Implementation
Implements a ticket end-to-end inside an isolated worktree. Commander spawns this for any ticket with code cha
Wh Builder
Build-phase agent. Executes exactly one task from plan.md using test-driven development inside an isolated git
Implementation Orchestrator
Use when the root orchestrator needs ONE phase of an arbitrado plan actually built — sequences test-writer (RE
Related Skills
Elite Build
Run the master build play — intent, human profile, gated stages, red proof, build, sniper tests, blind grade,
Harness Arena
Blind, LMSYS-style arena for comparing AI coding agent harnesses (Claude Code, Codex CLI, OpenClaw, Hermes, On
Harness Arena Fe
Blind, LMSYS-style arena for comparing AI coding agent harnesses (Claude Code, Codex CLI, OpenClaw, Hermes, On