DeepOps is a collection of automation tools for deploying and managing compute clusters that use NVIDIA GPUs with Kubernetes or Slurm. It supports environments ranging from full data-center clusters to individual machines needing GPU drivers, Docker, and NVIDIA container support.
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 NVIDIA/deepops --skill manage-model-cachegit clone --depth 1 https://github.com/NVIDIA/deepopsWrote 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/nvidia/deepops/manage-model-cache)<a href="https://agentmods.dev/skills/nvidia/deepops/manage-model-cache"><img src="https://agentmods.dev/badge/skills/nvidia/deepops/manage-model-cache/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/nvidia/deepops/manage-model-cache"><img src="https://agentmods.dev/badge/skills/nvidia/deepops/manage-model-cache.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00039 | $0.00903 |
| Opus 5.5 | $0.00016 | $0.00361 |
| Sonnet 5.5 | $0.00008 | $0.00181 |
| Haiku 4.5 | $0.00004 | $0.00090 |
Grade A, and why
manage-model-cache 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 11d 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 — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Manage a shared model cache
Run commands from the DeepOps repository root. This skill prepares model files, not inference services. Read the cache guide for the supported CLI version, exact configuration and working download example.
Preconditions
- Confirm the approved existing NFS mount, one designated writer account, reader group, model license/access, available capacity and immutable revision. Do not create mounts, change exports, create users, or grant access implicitly.
- Review
config/inventoryandconfig/group_vars/model-cache-warmer.yml. Exactly one host belongs tomodel-cache-warmer. All readers need matching numeric group IDs and an existing mount at the intended workload path. - Keep
HF_HOME,HF_TOKEN_PATHandHF_XET_CACHEprivate; share onlyHF_HUB_CACHE. Never share authentication storage or paste credentials into commands. - Use a private Python 3.10+ environment for
examples/model-cache/requirements.txt. Do not install into Ansible's or the serving framework's environment. No login or paid service is needed for the public example.
Procedure and gates
- Preview
ansible-playbook -i config/inventory playbooks/model-cache.yml --check --diff. Apply without--check --diffonly to the approved writer and cache path. The playbook must finish withfailed=0; repeat once to checkchanged=0. A successful playbook does not mean a model is cached. - As the writer, use
umask 0027, source the generatedenvironment.sh, and run the pinnedhf downloadwith a full commit hash and the approved file list as shown in the guide. Include every tokenizer/config file and weight shard required by the workload. Do not usemain,--local-dir, or deprecatedhf_transfer; do not have compute jobs warm the cache themselves. - Stop writes to the selected snapshot. As the actual reader UID, in the actual
workload mount context, run
python3 scripts/validation/validate_model_cache.pywith--json,--cache-dir,--repo-id,--revision, and one--require-fileper needed file. The guide has a complete command for the public example. - Accept only exit
0andok: true. Record revision,checked_filesandsnapshot_path. Pass that local path to the application and mount the entire cache read-only. The validator checks presence/readability, not checksums, complete model coverage, NFS locking, successful inference or GPU suitability. - For offline model loading, set
HF_HUB_OFFLINE=1only in the loading process. Do not apply it to a Hugging Face client calling a local inference server: it blocks that HTTP too. Offline mode is not a network access control.
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.
- 11d ago First seen · 62 lines · 39 tokens per session scan A ec9307a29863
manage-model-cache is a skill published in the GitHub repository NVIDIA/deepops (1,479 stars, last pushed today), licensed BSD-3-Clause. It adds 39 tokens to every session and 903 once invoked, about $0.0002 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 2026-10-07.
Other skills, from other repositories
tensorrt-llm
High-throughput LLM inference on NVIDIA GPUs.
agent-platform-tuning
Agent Platform Model Tuning. Use when you need to fine-tune open models or Gemini models using Agent Platform infrastructure. Don't use for model training outside Agent Platform, model deployment to endpoints (use agent-platform-deploy), or managing serving endpoints (use agent-platform-endpoint-management).
google-cloud-solution-guided-gke-ai-migration
Guides the migration of existing AI workloads (Cloud Run, Gemini API, Gemini Enterprise Agent Platform) to self-hosted GKE inference using gcloud and kubectl. Use when the user has an existing AI inference workload (on Cloud Run, the Gemini API, Gemini Enterprise Agent Platform, or a custom VM) and wants to move it to…
modal
Modal is a serverless cloud platform for running Python on demand, including on-demand GPUs. Use when deploying or serving AI/ML models, running GPU-accelerated workloads (training, fine-tuning, inference), serving web endpoints, scheduling batch jobs, or scaling Python code to cloud containers with the Modal SDK.
agent-platform-endpoint-management
Manages Agent Platform serving endpoints. Use when you need to create, list, describe, update, or delete serving endpoints for model deployment on Agent Platform. Also use when troubleshooting endpoint permission, quota, or resource busy errors. Don't use for deploying models to endpoints or for running model…
gke-inference
Deploys and optimizes AI/ML inference workloads on GKE, using GPUs, TPUs, and model servers. Use when deploying GKE inference servers, configuring GKE GPU resources for inference, or deploying LLMs on GKE. Don't use for migrating existing AI workloads to GKE (use google-cloud-solution-guided-gke-ai-migration), GKE RAG…