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 agentmods add skills/nvidia/tensorrt-model-connect/profile-modelnpx skills add NVIDIA/TensorRT-Model-Connect --skill profile-modelgit clone --depth 1 https://github.com/NVIDIA/TensorRT-Model-ConnectWhat 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 | $0.00047 | $0.01413 |
| Opus 5 | $0.00023 | $0.00707 |
| Sonnet 5 | $0.00009 | $0.00283 |
| Haiku 4.5 | $0.00005 | $0.00141 |
Grade A, and why
profile-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 2d 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 — 196 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Profile Model
Decide The Evidence Level
Choose one path before running:
| Question | Entry point | Claim boundary |
|---|---|---|
| Where is one model spending time? | tools/trtmc_profile.py |
diagnostic |
| Does a code change improve one owned workload? | profiler plus matching model testcase | local comparison |
| Is a model release-ready against its reference? | tools/perf_matrix.py |
release matrix |
| Does an optimized implementation qualify? | model-owned qualification producer | exact profile/target |
Profiler output is not automatically release or qualification evidence.
Preconditions And Provenance
Use the supported GPU/TensorRT environment and record:
git rev-parse HEAD
nvidia-smi --query-gpu=name,uuid,driver_version,pstate,power.draw \
--format=csv,noheader
python3 -c "import tensorrt as trt; print(trt.__version__)"
test -x ./build/trtmc
test -x ./build/trtmc-bench
Record the exact model revision, bundle SHA-256, native or optimized runtime path, effective config, target, warmups, timed iterations, inputs, token/sample counts, timing boundary, synchronization policy, and reference environment. Without those, label results exploratory.
If a team container is required:
./scripts/bootstrap_workspace.sh --id <team-id> \
--branch "$(git branch --show-current)" --detach
Correctness Before Timing
Select the owning model-first workload or E2E testcase and prove it passes before making performance claims:
PYTHONPATH=python:. python3 tools/trtmc_validate.py \
<model> <workload> \
--bundle <bundle.bundle> \
--output <validation-artifacts>
Do not time a candidate with a failed, skipped, or unrun comparison. Preserve the same model revision and workload when moving to profiling.
Quick Single-Model Diagnosis
PYTHONPATH=python:. python3 tools/trtmc_profile.py \
--model <model> \
--bundle <bundle.bundle> \
--prompt "<owned-testcase-prompt>" \
--max-new-tokens <N> \
--warmup 3 \
--iterations 10 \
--dtype float16 \
--trtmc-binary ./build/trtmc \
--hf-python <python> \
--json \
--output-dir <profile-dir>
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.
- 2d ago First seen · 196 lines · 47 tokens per session scan A c84b25ff46b6
profile-model is a skill published in the GitHub repository NVIDIA/TensorRT-Model-Connect (188 stars, last pushed 2d ago), licensed Apache-2.0. It adds 47 tokens to every session and 1,413 once invoked, about $0.0002 per session on Opus 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
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.
weights-and-biases
Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model registry with W&B - collaborative MLOps platform.
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.
perf-host-analysis
Analyze host/CPU overhead in TensorRT-LLM inference from nsys traces. Detect whether host overhead is the bottleneck using GPU idle ratio, host prep exposed ratio, and per-phase evidence. For regressions, isolate forward steps via allreduce/NVTX patterns, compare host operation breakdowns across versions, and identify…
perf-host-optimization
Profiles and optimizes TensorRT-LLM host/CPU overhead using lineprofiler (with nsys support planned). Runs iterative profile-analyze-optimize-validate rounds. Use when GPU utilization is low or optimizing PyExecutor throughput.
mlflow
Track ML experiments, manage model registry with versioning, deploy models to production, and reproduce experiments with MLflow - framework-agnostic ML lifecycle platform.