Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add magnus919/agent-skills --skill vllmgit clone --depth 1 https://github.com/magnus919/agent-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/magnus919/agent-skills/vllm)<a href="https://agentmods.dev/skills/magnus919/agent-skills/vllm"><img src="https://agentmods.dev/badge/skills/magnus919/agent-skills/vllm/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/magnus919/agent-skills/vllm"><img src="https://agentmods.dev/badge/skills/magnus919/agent-skills/vllm.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Privilege Escalation · line 63 Potential security issue detected. Manual review is recommended.Fix: Review the flagged content for security risks. Ensure no credentials, secrets, or sensitive data are exposed.
- medium Excessive Agency · line 35 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00196 | $0.03875 |
| Opus 5.5 | $0.00078 | $0.01550 |
| Sonnet 5.5 | $0.00039 | $0.00775 |
| Haiku 4.5 | $0.00020 | $0.00387 |
Grade A, and why
vllm scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 153 lines — stays where its author put it; the contents beside it link to each section on GitHub.
vLLM Inference Serving
Use this skill to operate vLLM as a production inference server: deploy it with Docker or Kubernetes, configure the model and engine (quantization, tensor parallelism, KV cache, context length), serve the OpenAI-compatible API surface, benchmark throughput and latency with comparable evidence, tune continuous batching, operate the GPUs underneath, and upgrade or roll back safely. This is a tool skill for one named engine. Serving methodology — engine selection, quantization trade-offs, deployment plans, regression triage — belongs to ml-engineering; local single-node GGUF serving with the llama.cpp stack belongs to llama-cpp. This skill owns the day-to-day operation of vLLM itself.
Operating contract
- Record the deployment before tuning it. Capture the vLLM version or image digest, model and revision, quantization, parallelism,
max-model-len, KV cache settings, batching limits, GPU inventory, and workload. The serving config template exists for exactly this. - Confirm the target, scope, and rollback path before acting. Read-only discovery (health probes,
/metrics,nvidia-smi) may proceed without confirmation. Mutations — restarting a server, changing serving args, scaling replicas, upgrading the image — require an explicit human directive naming the deployment. - A server that responds is not a server that serves.
/healthreturning 200 proves liveness, not that the model loaded or that inference works. Verify at the delivery boundary:/v1/modelsreports the served model and a representative request returns generated tokens. - Benchmark before and after every change. vLLM flags, defaults, and behavior change between releases; an unmeasured tuning change is a guess. Compare only matched conditions (version, model, GPU, context, batch, workload) and record the evidence in the benchmark run record.
- Keep evidence bounded. Summarize logs, configs, and metrics; never dump full server logs,
.envfiles, or HF tokens into chat.--enable-log-requestswith debug logging can leak prompt content; keep request logging off or redacted in shared sessions.
What ships with it
13 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- evals/evals.json 12 KB
- README.md 4.9 KB
- references/00-source-index.md 3.7 KB
- references/01-deployment.md 5.3 KB
- references/02-model-configuration.md 4.8 KB
- references/03-openai-api.md 3.2 KB
- references/04-benchmarking.md 3.8 KB
- references/05-batching-and-tuning.md 3.8 KB
- references/06-gpu-ops-and-lifecycle.md 4.9 KB
- scripts/vllm-health 7.2 KB
- templates/benchmark-run-record.md 2.4 KB
- templates/serving-config.md 2.6 KB
- tests/test_vllm_health.py 8.4 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 153 lines · 196 tokens per session scan A a90c5f21dc2b
vllm is a skill published in the GitHub repository magnus919/agent-skills (154 stars, last pushed today), licensed MIT. It adds 196 tokens to every session and 3,875 once invoked, about $0.0008 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
torchtitan
Pretrain LLMs at scale with PyTorch 4D parallelism.
modal
Serverless GPU cloud for ML jobs and model APIs.
tensorrt-llm
High-throughput LLM inference on NVIDIA GPUs.
lambda-labs
On-demand GPU cloud instances for ML training.
managed-model-endpoints
Register a model service in the managed family — a local model server container the daemon starts/stops on demand, or a remote upstream model API (https). Read the runbook, allocate a port (local only), compose idempotent start/stop scripts (local only), register once. Load when the user wants a model service…
map
Build and commit a Cortex function knowledge graph — maps structural dependencies and domain intent relationships across all AI functions in the project. Supports --reduce (default on) for transitive reduction of the dependency graph.