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/write-git-messagesnpx skills add NVIDIA/TensorRT-Model-Connect --skill write-git-messagesgit 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 | $0.00075 | $0.01503 |
| Opus 5 | $0.00037 | $0.00751 |
| Sonnet 5 | $0.00015 | $0.00301 |
| Haiku 4.5 | $0.00007 | $0.00150 |
Grade A, and why
write-git-messages 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 — 139 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Write Git Messages
Workflow
- Identify the artifact and exact comparison: commit, PR, squash/rebase message, or review response; base ref/SHA; head ref/SHA; and intended target.
- Inspect the actual change with
git status --short, `git diff --stat - Follow the repository's existing convention first. If it has no clear convention, default to Conventional Commits for commit and PR titles.
- Separate implemented behavior, static/CPU validation, target-hardware proof, model parity, package evidence, and performance/qualification. Do not promote one tier into another.
- Keep claims evidence-based. Do not invent issue numbers, benchmark results, tests, generated artifacts, reviewer decisions, or CI state.
- Return ready-to-use text followed by notes only when a limitation or choice matters.
For source rationale, read references/source-notes.md only when you need to explain or adjust the standard.
Commit Messages
Use this default shape:
type(scope): imperative summary
Why this change is needed and what behavior changes.
Mention important constraints, migrations, or risk.
Refs: #123
BREAKING CHANGE: explain incompatible behavior
Guidelines:
- Keep the first line short and self-contained; aim for 50 characters when practical.
- Separate the title from the body with a blank line.
- Use imperative mood in the summary:
add,fix,document,remove. - Use a meaningful type:
feat,fix,docs,test,refactor,perf,build,ci,chore, orrevert. - Add a scope only when it helps route ownership or understand impact.
- Put the "why" and user-visible behavior in the body; do not restate every file changed.
- Use footers for issue links, co-authors, and breaking changes.
- Do not add a co-author unless that person or tool actually authored part of the change and the user wants the attribution.
- Split unrelated work into separate commits when possible. If not possible, choose the dominant type and explain the combined scope in the body.
- Avoid vague titles such as
update files,misc fixes,work in progress, orfix stuff. - In this repo, never include
Claudein commit messages because the GitHub ruleset rejects it.
What ships with it
2 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.
- 2d ago First seen · 139 lines · 75 tokens per session scan A 2998d50ad1d8
write-git-messages is a skill published in the GitHub repository NVIDIA/TensorRT-Model-Connect (188 stars, last pushed 2d ago), licensed Apache-2.0. It adds 75 tokens to every session and 1,503 once invoked, about $0.0004 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.
git-workflow
Use when the task involves Git operations for the public xLLM repository, including choosing branch or tag names, preparing commits and pull requests, backporting fixes, checking repo-specific review expectations, or drafting commit messages from actual diffs.
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…
shep-kit:commit-pr
Use when ready to commit, push, and create a PR with CI verification. Triggers include "commit and pr", "push pr", "create pr", "ship it", or when implementation is complete and needs CI validation. Watches CI and auto-fixes failures. Part of the Shep autonomous SDLC platform — https://shep.bot.