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.
/plugin marketplace add EvilFreelancer/gpu-server-setupnpx agentmods add plugins/evilfreelancer/gpu-server-setup/marketplacegit clone --depth 1 https://github.com/EvilFreelancer/gpu-server-setupGrade A, and why
gpu-server-setup 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 yesterday.
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.
What it actually says
{
"name": "gpu-server-setup",
"owner": {
"name": "Pavel Rykov",
"email": "[email protected]"
},
"metadata": {
"description": "Prepare a Linux server with NVIDIA GPUs for LLM and neural-network workloads, from bare OS to a verified Docker + GPU base, with optional inference stacks and GPU power capping.",
"version": "0.2.1"
},
"plugins": [
{
"name": "gpu-server-setup",
"source": "./",
"description": "Prepare a Linux server with NVIDIA GPUs for LLM and neural-network workloads: NVIDIA driver + CUDA from the official repo, Docker Engine + NVIDIA Container Toolkit as the mandatory base, GPU passthrough into containers, staged verification gates, optional docker-compose presets for vLLM, Infinity, OpenWebUI and Ollama (llama.cpp via its official image), and GPU power-limit setup persisted via systemd.",
"version": "0.2.1",
"category": "ops",
"license": "MIT"
}
]
}
What ships with it
1 file beside marketplace.json 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.
- yesterday First seen · 22 lines scan A 27adcc5a8639
gpu-server-setup is a plugin published in the GitHub repository EvilFreelancer/gpu-server-setup (11 stars, last pushed 19d ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. 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 2026-08-30.
Other plugins, from other repositories
nvidia-cuopt-skills
Plugin marketplace listing 9 plugins: cuopt-user-rules, cuopt-developer, cuopt-install, cuopt-numerical-optimization-formulation, cuopt-numerical-optimization-api.
veloq
VeloQ is an agent-friendly profile-query CLI for Nsight Systems, Nsight Compute, and PyTorch/Kineto traces.
VeloQ
Agent-friendly GPU profile-query CLI.
flickies
Self-hosted video toolkit — lipsync, face restore, and ffmpeg ops (trim, concat, transcode, scale) via REST and MCP.
kernel-opt-agent
GPU kernel knowledge-base, benchmarking, profiling, and optimization-loop skills for CUDA, Triton, CuTe DSL, CUTLASS, PyTorch, and Nsight Compute workflows.
fmh66
Plugin marketplace listing 1 plugin: kernel-opt-agent.