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-TAO/tao-skill-bank --skill tao-run-on-brevgit clone --depth 1 https://github.com/NVIDIA-TAO/tao-skill-bankWrote 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-tao/tao-skill-bank/tao-run-on-brev)<a href="https://agentmods.dev/skills/nvidia-tao/tao-skill-bank/tao-run-on-brev"><img src="https://agentmods.dev/badge/skills/nvidia-tao/tao-skill-bank/tao-run-on-brev/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-tao/tao-skill-bank/tao-run-on-brev"><img src="https://agentmods.dev/badge/skills/nvidia-tao/tao-skill-bank/tao-run-on-brev.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 | $0.00106 | $0.02155 |
| Opus 5 | $0.00053 | $0.01077 |
| Sonnet 5 | $0.00021 | $0.00431 |
| Haiku 4.5 | $0.00011 | $0.00215 |
Grade C, and why
tao-run-on-brev scanned grade C with 2 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 8d 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.
Downloads and executes remote codehighSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
curl -fsSL https://raw.githubusercontent.com/brevdev/brev-cli/main/scripts/install-agent-skill.sh | bash Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -fsSL https://raw.githubusercontent.com/brevdev/brev-cli/main/scripts/install-agent-skill.sh | bash 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.
Brev — TAO execution glue
Standalone install? If this session was not initialized by the TAO skill bank plugin, run the
tao-setupskill first (host preflight, credentials, cross-skill discovery).
NVIDIA Brev provides on-demand GPU instances (pre-loaded with NVIDIA drivers,
CUDA, Docker, and the NVIDIA Container Toolkit). Brev is instance-based: you
provision an instance, run commands on it over brev exec, and delete it when
done.
This skill is deliberately thin. Provisioning and managing instances — create,
search by GPU/price, start/stop, delete, login — is owned by NVIDIA Brev's own
agent skill, not duplicated here. This skill covers only the TAO-specific part:
running a TAO container on a reached instance through the four-verb docker
contract, deferring the container-how to tao-run-on-docker over brev exec.
Provisioning: use the official Brev skill or MCP
NVIDIA Brev publishes an agent skill that manages instances in natural language ("create an A100 instance", "search for GPUs under $3/hr", "stop all my instances"). Install it once — it self-registers into your agent's skills dir and is discovered at runtime:
curl -fsSL https://raw.githubusercontent.com/brevdev/brev-cli/main/scripts/install-agent-skill.sh | bash
# installs to ~/.claude/skills/brev-cli/ , ~/.codex/skills/brev-cli/ , ~/.agents/skills/brev-cli/
Or connect the Brev MCP server (https://docs.nvidia.com/brev/_mcp/server).
Either one owns login/auth quirks, placement IDs, GPU search, and teardown flags.
It does not cover container execution on the instance — that is this skill.
Preflight for this skill: the brev CLI is on PATH and logged in (headless:
brev login --token "$BREV_API_TOKEN" before any other call), and you can reach a
target instance — poll with a two-word command until it succeeds before
issuing real work (a fresh instance reports RUNNING before sshd is up):
for i in $(seq 1 60); do brev exec <instance> "echo ok" 2>/dev/null | grep -qx ok && break; sleep 5; done
brev exec <instance> "echo ok" 2>/dev/null | grep -qx ok || { echo "instance not exec-ready"; exit 1; }
What ships with it
5 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.
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.
- 8d ago First seen · 153 lines · 106 tokens per session scan C fc0cc42415f8
tao-run-on-brev is a skill published in the GitHub repository NVIDIA-TAO/tao-skill-bank (88 stars, last pushed today), licensed Apache-2.0. It adds 106 tokens to every session and 2,155 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
modal-serverless-gpu
Run approved CPU or GPU work through OpenScience computejob on the user's configured Modal account. Use for isolated scientific scripts, dependency provisioning, durable outputs, logs, status, cancellation, and recovery. Never invoke the Modal SDK or CLI directly.
nvcf-self-managed-prerequisite
Install the prerequisites the NVCA operator / compute plane needs before nvcf-nvca-install can succeed: the operator tool nvcf-cli (required by the compute-plane stack's register-cluster step), KAI Scheduler (for the KAIScheduler feature gate), and the SMB CSI driver (for the sharedStorage Samba sidecar PVCs). The two…
nvca-chart-release
Release NVCA Operator chart changes from the native monorepo source to the vendored Helm chart. Use when updating the vendored NVCA Operator chart, changing NVCA image refs, publishing helm-nvca-operator, or validating the chart against a self-managed control plane.
nvca-values-customization
Customize NVCA Operator Helm chart values in the native monorepo. Use when modifying vendored defaults, changing stack-derived install values, adding deployment-time overrides, or updating scripts under deploy/helm/nvca-operator.
nvca-self-managed-install
Install or validate the NVCA Operator chart against a self-managed NVCF control plane from the native monorepo. Use when the control plane comes from deploy/stacks/self-managed and NVCA must be installed with stack-derived image repository settings.
cis-aws-compute-3.1
Ensure Amazon ECS task definitions using 'host' network mode do not allow privileged or root user access to the host.