benchmark-harness

80 / 100 · Low risk

Fine to try. Some of what it can do is expected: a coding skill usually runs build and test commands.

By anubhavg-icpl/vibe (GitHub) · coding · 13 stars · scanned at commit d51e7ac

Build a SWE-bench-style harness for a codebase with FAIL_TO_PASS / PASS_TO_PASS gating, contamination checks, and step-count metrics. Use when you need help with benchmark harness.

Works with any agent that supports SKILL.md skills.

What this skill can do

Runs commands on your computer

The author does not mention this. It runs automatically when the plugin is installed. Expected for a coding skill.

Found in 62 places.

.claude/settings.json, line 9"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" pre-bash'",
.claude/settings.json, line 19"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" pre-edit'",
.claude/settings.json, line 31"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" post-edit'",

Show 17 more

.claude/settings.json, line 41"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" post-bash'",
.claude/settings.json, line 52"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" route'",
.claude/settings.json, line 63"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" session-restore'",
.claude/settings.json, line 68"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/auto-memory-hook.mjs\" import'",
.claude/settings.json, line 79"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" session-end'",
.claude/settings.json, line 90"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/auto-memory-hook.mjs\" sync'",
.claude/settings.json, line 102"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" compact-manual'"
.claude/settings.json, line 106"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" session-end'",
.claude/settings.json, line 116"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" compact-auto'"
.claude/settings.json, line 120"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" session-end'",
.claude/settings.json, line 131"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" status'",
.claude/settings.json, line 142"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" post-task'",
.claude/settings.json, line 153"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/hook-handler.cjs\" notify'",
.claude/settings.json, line 162"command": "sh -c 'exec node \"${CLAUDE_PROJECT_DIR:-.}/.claude/helpers/statusline.cjs\"'"
.mcp.json, line 4"command": "npx",
hooks/cmods/README.md, line 64"os.system("
hooks/cmods/README.md, line 65"subprocess.call.*shell=True"

and 42 more places

Downloads and installs other software

The author does not mention this. It runs automatically when the plugin is installed.

Found in 10 places.

.claude/settings.json, line 166"Bash(npx @claude-flow*)",
.claude/settings.json, line 167"Bash(npx claude-flow*)",
.mcp.json, line 4"command": "npx",

Show 7 more

hooks/cmods/README.md, line 96npx eslint --fix "$FILE" 2>/dev/null
hooks/cmods/post-edit-format.sh, line 29npx prettier --write "$FILE" 2>/dev/null && FORMATTED=true
hooks/cmods/post-edit-format.sh, line 32npx biome format --write "$FILE" 2>/dev/null && FORMATTED=true
hooks/cmods/post-edit-format.sh, line 80npx prettier --write "$FILE" 2>/dev/null && FORMATTED=true
hooks/cmods/post-edit-format.sh, line 86npx prettier --write "$FILE" 2>/dev/null && FORMATTED=true
hooks/cmods/pre-commit-lint.sh, line 38echo "$files" | xargs npx eslint --max-warnings 0 2>/dev/null
hooks/hooks.json, line 106"command": "#!/bin/bash\n# Run TypeScript check after editing TS files\ninput=$(cat)\nfile_path=$(echo \"$input\" | jq -r '.tool_input.file_path // \"\"')\n\nif

Install

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

if [ -e ~/.agents/skills/'benchmark-harness' ] || [ -L ~/.agents/skills/'benchmark-harness' ]; then echo 'A skill folder named benchmark-harness already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/anubhavg-icpl/vibe' "$d" && git -C "$d" checkout --quiet 'd51e7ac9a34111cd2e63daed83c2198f91570050' -- 'skills/benchmark-harness' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/benchmark-harness' ~/.agents/skills/'benchmark-harness' && 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/'benchmark-harness' ~/.claude/skills/'benchmark-harness' (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 anubhavg-icpl/vibe shows the score of its lowest-scoring skill.

SkillTrust badge for anubhavg-icpl/vibe

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/anubhavg-icpl/vibe.svg)](https://zalavijaysinh.github.io/skilltrust/repos/anubhavg-icpl/vibe/)

Score history

Date	Version	Score
2026-11-08	`d51e7ac`	80