hodamousavipour

Eval Grader — Development agent for Claude Code

Development community

Use when an eval case's subject run has finished and needs an independent verdict against its expectations.

How to install Eval Grader

Installs to ~/.claude/agents/hodamousavipour-claude-code-llm-engineering-eval-grader.md

Terminal

mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/hodamousavipour/claude-code-llm-engineering/HEAD/agents/eval-grader.md -o ~/.claude/agents/hodamousavipour-claude-code-llm-engineering-eval-grader.md

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

What Eval Grader does

name: eval-grader description: Use when an eval case's subject run has finished and needs an independent verdict against its expectations. Spawned by scripts/run_eval.py via `claude -p --agent eval-grader` after each subject run, non-interactively, on a model pinned distinct from the subject's — never invoke this agent to modify the code, skill, or transcript under review, only to judge it. model: opus tools:

Read
Grep
Glob

You are an independent judge grading a completed ev

Alternatives in Development

Sales Automator — Draft cold emails, follow-ups, and proposal templates 2.8k ★
Engram Assessor — Independent grader of learner productions for the Engram learning plugin 1.4k ★
Eval Implementer — Implementer role of the D365FO agent eval loop 136 ★