DonthiShivani

Agent Eval Engineer — Design agent for Claude Code

Design community

Use to design and run regression evaluations for an agent, skill, tool, or prompt — happy-path, edge-case, adversarial, and historical-regression cases — and to grade results independently of whoever.

How to install Agent Eval Engineer

Installs to ~/.claude/agents/donthishivani-iva-agent-eval-engineer.md

Terminal

mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/DonthiShivani/iva/HEAD/agents/agent-eval-engineer.md -o ~/.claude/agents/donthishivani-iva-agent-eval-engineer.md

Restart Claude Code, or start a new session, for it to be picked up.

What Agent Eval Engineer does

name: agent-eval-engineer description: Use to design and run regression evaluations for an agent, skill, tool, or prompt — happy-path, edge-case, adversarial, and historical-regression cases — and to grade results independently of whoever built what's being evaluated. Use whenever a project has its own agents, tools, or prompts and needs evidence they still work before a change ships, not just a manual spot-check. tools: Read, Grep, Glob, Write, Bash maxTurns: 50

You are an independent

Alternatives in Design

Workflow Architect Agent — Workflow design specialist who maps complete workflow trees for every system, user journey, and agent interact 669 ★
Canvas Designer — Use this agent for UI/UX design and implementation — production-grade interfaces with intentional aesthetic 523 ★
Asset Producer — Produces one assigned visual asset production unit for the asset stage 517 ★