← All skills

vlm_judge_claude

95 / 100 · Low risk

Fine to try.

By nssmd/RoboRSI (GitHub) · media · 179 stars · scanned at commit 90c8569

VLM-as-judge using a separate Claude CLI process. Spawns `claude -p --bare --output-format json` with image paths in the prompt; the subprocess Reads the images and returns a strict JSON success verdict. Used by atomic.judge sub-skills and long_horizon.progress_judge to replace sim-side success_predicate.

Works with any agent that supports SKILL.md skills.

What this skill can do

Install

Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.

if [ -e ~/.agents/skills/'vlm-judge-claude' ] || [ -L ~/.agents/skills/'vlm-judge-claude' ]; then echo 'A skill folder named vlm-judge-claude already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/nssmd/RoboRSI' "$d" && git -C "$d" checkout --quiet '90c8569705bf4004a4f79e34e650c87ab7f075d4' -- 'roborsi/embodied/skills/_lib/judging/vlm_judge_claude' && mkdir -p ~/.agents/skills && cp -R "$d"/'roborsi/embodied/skills/_lib/judging/vlm_judge_claude' ~/.agents/skills/'vlm-judge-claude' && rm -rf "$d"; fi

Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'vlm-judge-claude' ~/.claude/skills/'vlm-judge-claude' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.

Badge for your README

The badge for nssmd/RoboRSI shows the score of its lowest-scoring skill.

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/nssmd/RoboRSI.svg)](https://zalavijaysinh.github.io/skilltrust/repos/nssmd/RoboRSI/)

Score history