NVIDIA/TensorRT-Model-Connect is a collection of C++ reference implementations for deploying supported Hugging Face models with NVIDIA TensorRT, a system that runs trained models to produce inference results. It is for developers who want to build and run supported models or evaluate model integrations through TensorRT. The catalogue entries provide skills and instructions for using this model deployment workflow.
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/TensorRT-Model-Connect --skill transform-modelgit clone --depth 1 https://github.com/NVIDIA/TensorRT-Model-ConnectWrote 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/tensorrt-model-connect/transform-model)<a href="https://agentmods.dev/skills/nvidia/tensorrt-model-connect/transform-model"><img src="https://agentmods.dev/badge/skills/nvidia/tensorrt-model-connect/transform-model/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/tensorrt-model-connect/transform-model"><img src="https://agentmods.dev/badge/skills/nvidia/tensorrt-model-connect/transform-model.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00033 | $0.00511 |
| Opus 5.5 | $0.00013 | $0.00204 |
| Sonnet 5.5 | $0.00007 | $0.00102 |
| Haiku 4.5 | $0.00003 | $0.00051 |
Grade A, and why
transform-model 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 — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Transform a Model
Define the checkpoint, revision, task, target hardware, requested precision, and required evidence. Read the upstream config and reference implementation, then inspect the closest family only as a pattern.
Extend an owner only when checkpoint identity, weight mapping, graph dataflow,
native request lifecycle, and correctness contract genuinely match. Otherwise
create one new families/<family>/ directory.
Implement one vertical slice
support.pyrecognizes exact model metadata and declares supported tasks plus one meaningful default. Zero or multiple owners are errors.model.pyexposes one plainbuild(request, writer)function. It owns model config, weights, TensorRT graphs, precision, parallelism, and bundle section semantics. It must not inherit a builder or import another family.runtime/CMakeLists.txtbuildstrtmc_model_<family>. The family factory and pipeline implement the required abstract interface fromtrtmc/task.hand drive engines through the public backend contract.requirements.txtcontains only optional packages needed by this family.tests/owns direct build/runtime E2E, manifests, thresholds, fixtures, and focused Python or C++ tests.
Keep model behavior inside the owner even when another family looks similar. Do not modify core merely to reduce duplicated lines. If the requested user behavior cannot be expressed by an existing Task interface, stop and describe the required shared contract before expanding scope.
Prove the smallest real path
First make one exact testcase work end to end:
checkpoint -> support resolution -> family build -> bundle
-> family DSO -> abstract Task API -> real output
Then run repository ownership checks and the selected family tests:
PYTHONPATH=core/builder:apps/benchmark:. python3 -m tools.model_ci validate
PYTHONPATH=core/builder:apps/benchmark:. python3 -m pytest \
families/<family>/tests --e2e-model <family> -q
git diff --check
What ships with it
1 file 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.
- 1mo ago Changed · -139 lines · -30 tokens per session a34cb9bfabda
- 1mo ago First seen · 197 lines · 63 tokens per session scan A 8d2666374edd
transform-model is a skill published in the GitHub repository NVIDIA/TensorRT-Model-Connect (276 stars, last pushed today), licensed Apache-2.0. It adds 33 tokens to every session and 511 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
cutedsl_megamoe
Skill "cutedsl_megamoe" from flashinfer-ai/flashinfer, covering updating the cutedsl megamoe kernel src, layout, when the kernel team drops a new version of src/ and what not to update here.
huggingface-transformers
Use when running or fine-tuning Hugging Face Transformers models in Python — pipeline() inference for text classification and generation, loading weights with AutoModel/AutoTokenizer, devicemap and dtype memory management, tokenizer usage, Trainer fine-tuning, and BGE embeddings for Chinese RAG. Covers transformers…
graphsignal-profiler
Set up the Graphsignal Profiler for inference workloads — vLLM, SGLang, PyTorch, and dstack services. Use when the user wants GPU profiling, tracing, or monitoring for inference, asks about graphsignal-run or graphsignal.watch(), or asks about CUPTI / Prometheus / OTLP setup.
huggingface-accelerate
Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.
mlflow
Track ML experiments, manage model registry with versioning, deploy models to production, and reproduce experiments with MLflow - framework-agnostic ML lifecycle platform.
dup-classes
Verify whether generated Java classes duplicate openai-java models by comparing fields/types (names may differ). Use when checking for duplicate model coverage.