tensorrt-llm

Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs (A100/H100), when you need 10-100x faster inference than PyTorch, or for serving models with quantization (FP8/INT4), in-flight batching, and multi-GPU scaling.

Quality

86%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

72%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable with executable code throughout and uses progressive disclosure well, pointing to real reference files. It loses points for promotional/encyclopedic padding and for missing validation checkpoints in deployment workflows.

Suggestions

Remove the promotional 'Performance benchmarks' section and the 'Supported models' list, or compress them to a single line — Claude already knows these models and the throughput claims add tokens without actionability.

Add validation checkpoints to the serving and multi-GPU workflows (e.g., confirm the trtllm-serve process is listening on :8000 and check that all tensor-parallel ranks initialized before sending requests).

Delete the one-line intro 'NVIDIA's open-source library for optimizing LLM inference...' since it duplicates the frontmatter description.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient and code-heavy, but includes padding that does not earn its place: the intro line restates the description, the 'Performance benchmarks' section is promotional ('100× faster'), and the 'Supported models' list enumerates models Claude already knows about.	2 / 3
Actionability	Provides fully executable, copy-paste-ready guidance across install, basic inference, serving (with curl), FP8 quantization, and multi-GPU configs — specific commands and code with no pseudocode.	3 / 3
Workflow Clarity	Quick start sequences install → inference → serving reasonably, but there are no validation checkpoints (e.g., confirming the server compiled/started, verifying multi-GPU rank setup) for what are fragile deployment operations, capping clarity at 2.	2 / 3
Progressive Disclosure	Clear overview with well-signaled one-level-deep references to optimization.md, multi-gpu.md, and serving.md — all three files exist in ./references/ — with content appropriately split and easy to navigate.	3 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, well-triggered, and clearly distinguishes this skill from adjacent inference tools, explicitly covering both what it does and when to use it. It is concise without padding or over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities — 'maximum throughput and lowest latency', 'quantization (FP8/INT4)', 'in-flight batching', and 'multi-GPU scaling' — in third person voice, matching the 'lists multiple specific concrete actions' anchor.	3 / 3
Completeness	Explicitly answers both what ('Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency') and when ('Use for production deployment... when you need 10-100x faster inference... or for serving models with quantization'), with an explicit 'Use for...' trigger clause.	3 / 3
Trigger Term Quality	Covers natural terms a user would actually say for this domain — 'LLM inference', 'production deployment on NVIDIA GPUs', 'quantization (FP8/INT4)', 'in-flight batching', 'multi-GPU' — giving good coverage of likely trigger phrases.	3 / 3
Distinctiveness Conflict Risk	The niche is sharply defined — NVIDIA TensorRT on A100/H100 GPUs for production LLM serving — making it unlikely to trigger for unrelated skills and clearly distinct from generic ML/inference tools.	3 / 3
	Total	12 / 12 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: NousResearch/hermes-agent
Path: optional-skills/mlops/tensorrt-llm/SKILL.md
Commit: 4c9628e

Reviewed: about 13 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.