← All skills
vllm-bench-serve
79 / 100 · Review before installing
Check before installing. Some of what it can do is expected: a devops skill usually runs commands, installs tools and uses cloud credentials.
By ascend-ai-coding/awesome-ascend-skills (GitHub) · devops · 184 stars · scanned at commit 75ec89c
Interactive online benchmark orchestrator for vLLM inference services using `vllm bench serve`. Supports single benchmarks, multi-case batch execution with result aggregation, and auto-optimization to find optimal concurrency/throughput under latency SLO constraints (TTFT, TPOT, P99, success rate). Use this skill whenever the user wants to benchmark, stress test, or measure performance of a running LLM/multimodal/embedding inference service — even if they don't say "vllm bench serve" explicitly…
Works with any agent that supports SKILL.md skills.
What this skill can do
Sends data to these websites: ip, target
This is disclosed by the author in the skill's SKILL.md or README. Expected for a devops skill.
Found in 15 places.
Show 12 more
- skills/inference/vllm-bench-serve/references/environment-checks.md, line 46
curl -s --connect-timeout 5 http://TARGET/health - skills/inference/vllm-bench-serve/references/troubleshooting.md, line 71
1. Verify the service is running: `curl -s http://IP:PORT/health` - skills/inference/vllm-bench-serve/references/troubleshooting.md, line 83
- Wrong `--served-model-name`: the model name in requests doesn't match what the service expects. Check with `curl http://IP:PORT/v1/models`. - skills/inference/vllm-bench-serve/references/troubleshooting.md, line 88
- Verify model name: `curl -s http://IP:PORT/v1/models | python3 -c "import sys,json; [print(m['id']) for m in json.load(sys.stdin)['data']]"` - skills/inference/vllm-bench-serve/scripts/check_bench_env.sh, line 55
AIOHTTP_VER=$(python3 -c "import aiohttp; print(aiohttp.__version__)" 2>/dev/null) && \ - skills/inference/vllm-bench-serve/scripts/check_bench_env.sh, line 62
HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout 5 "$BASE_URL/health" 2>/dev/null) || HTTP_CODE="000" - skills/inference/vllm-bench-serve/scripts/probe_service.sh, line 31
HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout "$TIMEOUT" "$BASE_URL/health" 2>/dev/null) || HTTP_CODE="000" - skills/inference/vllm-bench-serve/scripts/probe_service.sh, line 37
MODELS_RAW=$(curl -s --connect-timeout "$TIMEOUT" "$BASE_URL/v1/models" 2>/dev/null) || MODELS_RAW="" - skills/inference/vllm-bench-serve/scripts/probe_service.sh, line 57
COMP_CODE=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout "$TIMEOUT" \ - skills/inference/vllm-bench-serve/scripts/probe_service.sh, line 66
CHAT_CODE=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout "$TIMEOUT" \ - skills/inference/vllm-bench-serve/scripts/probe_service.sh, line 75
EMB_CODE=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout "$TIMEOUT" \ - skills/inference/vllm-bench-serve/scripts/probe_service.sh, line 84
RERANK_CODE=$(curl -s -o /dev/null -w "%{http_code}" --connect-timeout "$TIMEOUT" \
Runs commands on your computer
This is disclosed by the author in the skill's SKILL.md or README. Expected for a devops skill.
Found in 12 places.
Show 9 more
- skills/inference/vllm-bench-serve/scripts/auto_optimize.py, line 62
result = subprocess.run( - skills/inference/vllm-bench-serve/scripts/auto_optimize.py, line 63
cmd, shell=True, capture_output=True, text=True, timeout=600 - skills/inference/vllm-bench-serve/scripts/auto_optimize.py, line 69
except subprocess.TimeoutExpired: - skills/inference/vllm-bench-serve/scripts/run_batch.sh, line 104
timeout "$CASE_TIMEOUT" bash -c "$FULL_CMD" > "$RESULT_DIR/${CASE_NAME}.log" 2>&1 || _EC=$? - skills/inference/vllm-bench-serve/scripts/run_batch.sh, line 106
bash -c "$FULL_CMD" > "$RESULT_DIR/${CASE_NAME}.log" 2>&1 || _EC=$? - skills/inference/vllm-bench-serve/scripts/run_batch.sh, line 119
timeout "$CASE_TIMEOUT" bash -c "$FULL_CMD" 2>&1 | tee "$RESULT_DIR/${CASE_NAME}.log" || EXIT_CODE=$? - skills/inference/vllm-bench-serve/scripts/run_batch.sh, line 124
bash -c "$FULL_CMD" 2>&1 | tee "$RESULT_DIR/${CASE_NAME}.log" || EXIT_CODE=$? - skills/inference/vllm-bench-serve/scripts/run_bench.sh, line 56
timeout "$BENCH_TIMEOUT" bash -c "$BENCH_CMD" 2>&1 | tee "$LOG_FILE" || EXIT_CODE=$? - skills/inference/vllm-bench-serve/scripts/run_bench.sh, line 61
bash -c "$BENCH_CMD" 2>&1 | tee "$LOG_FILE" || EXIT_CODE=$?
Downloads and installs other software
The author does not mention this. Expected for a devops skill.
Found in 1 place.
Safer alternatives
Low-risk skills in the same category (devops) that look related. Check that one does what you need.
Install
Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.
if [ -e ~/.agents/skills/'vllm-bench-serve' ] || [ -L ~/.agents/skills/'vllm-bench-serve' ]; then echo 'A skill folder named vllm-bench-serve already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/ascend-ai-coding/awesome-ascend-skills' "$d" && git -C "$d" checkout --quiet '75ec89ce9d401be8ac95980e1c3f42f8d4315057' -- 'skills/inference/vllm-bench-serve' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/inference/vllm-bench-serve' ~/.agents/skills/'vllm-bench-serve' && 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/'vllm-bench-serve' ~/.claude/skills/'vllm-bench-serve' (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 ascend-ai-coding/awesome-ascend-skills shows the score of its lowest-scoring skill.
[](https://zalavijaysinh.github.io/skilltrust/repos/ascend-ai-coding/awesome-ascend-skills/)
Score history
| Date | Version | Score |
|---|
| 2026-11-08 | 75ec89c | 100 |