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/submit-github-prnpx skills add NVIDIA/TensorRT-Model-Connect --skill submit-github-prgit 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 | $0.00058 | $0.01591 |
| Opus 5 | $0.00029 | $0.00796 |
| Sonnet 5 | $0.00012 | $0.00318 |
| Haiku 4.5 | $0.00006 | $0.00159 |
Grade A, and why
submit-github-pr 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.
How it starts
The opening of the file, as written. The whole thing — 189 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Submit GitHub PR
Boundaries
- Active repository:
NVIDIA/TensorRT-Model-Connect. - Use the local
githubremote and target GitHubmain. - Never push directly to
main. - Publish only the change the user placed in scope.
- Open a draft PR by default; omit
--draftonly when the user explicitly asks for a ready-for-review PR and the evidence supports that state. - Do not add labels, reviewers, milestones, projects, or auto-merge unless the user explicitly requests them. The authorized Internal CI trigger is the exception described below.
- Do not merge from this skill. Use
$pr-babysitterfor monitoring and merge. - Use
$write-git-messagesfor commit, PR, squash, and rebase text.
1. Authenticate And Verify The Remote
gh auth status
gh repo view NVIDIA/TensorRT-Model-Connect \
--json nameWithOwner,url,defaultBranchRef
git remote get-url github
git fetch github main
The remote URL must resolve to
https://github.com/NVIDIA/TensorRT-Model-Connect.git or its authorized SSH
equivalent. If gh cannot read the private repository, stop for safe
reauthentication. Do not inspect or replay stored credentials.
2. Confirm Branch And Exact Scope
BRANCH=$(git branch --show-current)
test -n "$BRANCH"
test "$BRANCH" != main
BASE_SHA=$(git rev-parse github/main)
HEAD_SHA=$(git rev-parse HEAD)
git status --short --branch
git merge-base --is-ancestor "$BASE_SHA" "$HEAD_SHA"
git diff --stat "$BASE_SHA"...HEAD
git diff --name-status "$BASE_SHA"...HEAD
git log --oneline "$BASE_SHA"..HEAD
Review focused diffs, staged changes, and untracked files. Do not publish:
- unrelated user changes;
- generated artifacts or secrets not intended for review;
- an empty branch;
- a branch based on the wrong repository or target;
- commit messages containing
Claude.
If the branch is behind or diverged from current github/main, rebase only when
that history rewrite is within the user's request. Revalidate after any rebase.
3. Validate The Publishable Head
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.
- yesterday First seen · 189 lines · 58 tokens per session scan A 6122d9afd470
submit-github-pr is a skill published in the GitHub repository NVIDIA/TensorRT-Model-Connect (188 stars, last pushed yesterday), licensed Apache-2.0. It adds 58 tokens to every session and 1,591 once invoked, about $0.0003 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.
mlflow
Track ML experiments, manage model registry with versioning, deploy models to production, and reproduce experiments with MLflow - framework-agnostic ML lifecycle platform.
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.