llm-benchmark
Use when the user wants to test, compare, or choose local Ollama models for their machine.
Skill by KerberosClaw
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/kerberosclaw-llm-benchmark
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: documents, backend, testing, ml
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 26, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/llm-benchmark- Your home directory (~)
What's inside
- Skill name
llm-benchmark- Description
- Use when the user wants to test, compare, or choose local Ollama models for their machine. Checks Ollama/GPU state, recommends model sizes from available VRAM, preserves existing benchmark records, pulls only approved models, runs repeatable benchmarks, restores stopped services, and writes a markdown comparison report. NOT for hosted API model evaluation or subjective chat-quality judging without local benchmark commands.
- License
- MIT
- Instructions
- 189 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
- Instructions call the network
Files (2)
- SKILL.md
- benchmark.py · scripts/
SKILL.md (first lines)
# LLM Benchmark Skill
You are a local LLM benchmarking specialist. 你負責用可重現的命令測 Ollama 模型效能,保護使用者現有服務狀態,並把推薦和限制講清楚。
## 不適用
- 不評測雲端 API 模型。
- 不用單次主觀聊天感覺取代 benchmark。
- 不在未確認硬體與服務狀態前直接 pull 大模型或重啟服務。
執行以下完整流程:
## Step 0:環境檢查
```bash
# 確認 Ollama 是否安裝並運行
curl -s http://localhost:11434/api/version 2>/dev/null || echo "NOT_RUNNING"
which ollama 2>/dev/null || echo "NOT_INSTALLED"
```
- 若 **未安裝**:執行 `curl -fsSL https://ollama.com/install.sh | sh`,再啟動服務
- 若 **未運行**:執行 `ollama serve &` 或 `systemctl start ollama`
- 確認成功後繼續
## Step 0.5:VRAM 清空(benchmark 前必做)
### 1. 停止其他吃 GPU / RAM 的本地 LLM 服務(若有且正在運行)
```bash
# 檢查是否有 LLM gateway 類容器正在運行(例:OpenClaw、LM 代理服務)
LLM_GATEWAY="your-llm-gateway" # 換成你的服務名,例:openclaw
docker ps --format '{{.Names}}' | grep -i "$LLM_GATEWAY"
```
- 若有相關容器正在運行 → 停止它:
```bash
LLM_STACK_DIR="$HOME/your-llm-stack" # 換成你的 compose 目錄
GATEWAY_SERVICE="gateway" # 換成你的服務名
cd "$LLM_STACK_DIR" && docker compose stop "$GATEWAY_SERVICE"
```
- 若無相關容器,或容器已停止 → 跳過此步驟
記錄該服務是否原本是啟動的,**benchmark 完成後需還原狀態**。
### 2. 重啟 Ollama 清除 VRAM(必做)
```bash
sudo systemctl restart ollama
sleep 5
```
### 3. 確認 VRAM 已釋放
```bash
nvidia-smi --query-gpu=memory.used,memory.free --format=csv,noheader,nounits
```
確認 `memory.used` 降到 ~1500 MB 以下再繼續。若仍偏高,再等 5 秒重確認。
### 4. Benchmark 完成後還原服務
若步驟 1 有停止服務,benchmark 全部完成後執行:
```bash
cd "$LLM_STACK_DIR" && docker compose start "$GATEWAY_SERVICE"
```
---
## Step 1:取得 GPU/VRAM 資訊
```bash
nvidia-smi --query-gpu=name,memory.total,memory.free --format=csv,noheader,nounits 2>/dev/null \
|| rocm-smi --showmeminfo vram 2>/dev/null \
|| echo "NO_GPU"
```
依據 **可用 VRAM** 推薦模型最大參數量(以 Q4 量化為基準):
| 可用 VRAM | 推薦最大 B |
|-----------|-----------|
| < 4 GB | 3B |
| 4–6 GB | 7B |
| 6–8 GB | 8B |
| 8–10 GB | 12–14B |
| 10–14 GB | 14B |
| 14–16 GB | 14–20B |
| > 16 GB | 30B+ |
**向使用者說明推薦理由**,列出推薦模型清單(附 Ollama model tag)。
## Step 2:確認要測試的模型
詢問使用者確認測試清單(若使用者已在指令中指定則略過)。
同時列出機器上 **現有模型**:
```bash
curl -s http://localhost:11434/api/tags | python3 -c "
import sys,json
data=json.load(sys.stdin)
for m in data.get('models',[]):
print(m['name'], round(m['size']/1024/1024), 'MB')
"
```
## Step 3:檢查既有 Benchmark 記錄
檢查 `~/benchmark_results.json` 是否存在:
```bash
python3 -c "
import json, os
path = os.path.expanduser('~/benchmark_results.json')
if not os.path.exists(path):
print('NO_BENCHMARK_FILE')
else:
data = json.load(open(path))
benchmarked = [e['model'] for e in data]
print('BENCHMARKED:', benchmarked)
"From KerberosClaw/kc_ai_skills, MIT.
More from KerberosClaw/kc_ai_skills
- 0.676score
align
KerberosClaw
Reconcile existing repository documentation with the current implementation, especially after verified hotfixes or after QA in an unattended task.
SkillAFreshness: A★ 79Claude Code +9 - 0.676score
ctf-kit
KerberosClaw
Use when the user is solving an authorized CTF / lab reverse-engineering challenge focused on Windows application authentication or license-check bypass.
SkillAFreshness: A★ 79Claude Code - 0.676score
goal-engineer
KerberosClaw
Author a self-contained unattended dispatch for a generate-and-select loop (generate candidates, grade, iterate, keep the best for human selection), or for a…
SkillAFreshness: A★ 79Claude Code +9 - 0.671score
workflow-router
KerberosClaw
Use when the user is unsure which kc_ai_skills workflow skill to use, or describes work involving PRD, SD/software design, FR, AC/acceptance criteria, ADR…
SkillAFreshness: A★ 79Claude Code +9 - 0.671score
adr
KerberosClaw
Use when a durable technical decision was just made in conversation and should be recorded, or when user asks to write/record an ADR (開 ADR / 記個決策 / 這要不要 ADR).
SkillAFreshness: A★ 79Claude Code +9 - 0.651score
spec
KerberosClaw
Use when the user wants a spec-driven development workflow for implementing a feature in the current codebase, from fuzzy idea or existing active spec through…
SkillAFreshness: A★ 79Claude Code +9 - 0.647score
ios-resign
KerberosClaw
Use when the user wants to rebuild, re-sign and reinstall one of their own iOS apps onto their iPhone(s) — typically because a free Apple ID provisioning…
SkillAFreshness: A★ 79Claude Code +9 - 0.632score
wrap-up
KerberosClaw
Use ONLY when the user's latest message explicitly contains the `/wrap-up` command.
SkillBFreshness: B★ 79Claude Code +9 - 0.631score
prep-repo
KerberosClaw
Prepare an existing project for GitHub or public release with scope-appropriate checks for sensitive data, documentation, installation, tests and release…
SkillBFreshness: B★ 79Claude Code +9 - 0.631score
diagnose
KerberosClaw
Use when the user reports misbehaving software to investigate — a bug, crash, wrong output, flaky behavior, or '為什麼壞掉/不會動/查一下' — and the root cause is not yet…
SkillBFreshness: B★ 79Claude Code +9 - 0.631score
job-scout
KerberosClaw
Use when the user is considering applying to, interviewing with, or accepting an offer from a company and wants due diligence on the company and optionally a…
SkillBFreshness: B★ 79Claude Code +9 - 0.611score
memory-lint
KerberosClaw
Use when the user wants to lint a Claude Code memory directory (~/.claude/memory or custom path) for index inconsistency, broken cross-links, stale project…
SkillBFreshness: B★ 79Claude Code +9
Signals
- Stars
- 79
- Forks
- 14
- Open issues
- 0
- Last commit (this folder)
- Jul 30, 2026 (3 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/KerberosClaw/kc_ai_skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from KerberosClaw
- 0.676score
ctf-kit
KerberosClaw
Use when the user is solving an authorized CTF / lab reverse-engineering challenge focused on Windows application authentication or license-check bypass.
SkillAFreshness: A★ 79Claude Code - 0.676score
align
KerberosClaw
Reconcile existing repository documentation with the current implementation, especially after verified hotfixes or after QA in an unattended task.
SkillAFreshness: A★ 79Claude Code +9 - 0.676score
goal-engineer
KerberosClaw
Author a self-contained unattended dispatch for a generate-and-select loop (generate candidates, grade, iterate, keep the best for human selection), or for a…
SkillAFreshness: A★ 79Claude Code +9 - 0.671score
workflow-router
KerberosClaw
Use when the user is unsure which kc_ai_skills workflow skill to use, or describes work involving PRD, SD/software design, FR, AC/acceptance criteria, ADR…
SkillAFreshness: A★ 79Claude Code +9 - 0.671score
adr
KerberosClaw
Use when a durable technical decision was just made in conversation and should be recorded, or when user asks to write/record an ADR (開 ADR / 記個決策 / 這要不要 ADR).
SkillAFreshness: A★ 79Claude Code +9 - 0.651score
spec
KerberosClaw
Use when the user wants a spec-driven development workflow for implementing a feature in the current codebase, from fuzzy idea or existing active spec through…
SkillAFreshness: A★ 79Claude Code +9
Cite this page
llm-benchmark skill. Agent Mart. https://agentmart.dev/skills/kerberosclaw-llm-benchmark (last updated 2026-10-24)