perf-host-optimization
perf-host-optimization is an MLOps & model serving skill for AI agents, published by NVIDIA in the TensorRT-LLM repository on GitHub. Skillspace rates perf-host-optimization C (77/100) and it passed the safety scan; a judge model has also reviewed it.
Profiles and optimizes TensorRT-LLM host/CPU overhead using line_profiler (with nsys support planned). Runs iterative profile-analyze-optimize-validate rounds. Use when GPU utilization is low or optimizing PyExecutor throughput.
- Grade
- C
- Quality
- 77 / 100
- Safety
- Passed
- GitHub stars
- 14,127
Placed #3 of 6 in the judged category final for MLOps & model serving.
Not launch-quality. It is graded C with a clean safety verdict, but our quality filters exclude it: a model that read it found it written only for the repository it lives in. So it is not in our launch-quality set, the skills our MCP server searches and recommends. How launch-quality is decided
Our two independent classifiers differ on this skill's category, so treat the placement as provisional.
This rating describes the version of the repository we crawled; it may have changed since, so go by its current files.
Our review
- Architecture
- 94 / 100
- Maturity
- Packaged: licensed, with a version or references
- Security scan
- No issues found
- License
- Not declared. Check the repository before reusing it.
- Adoption
- Adopted by 3 other repositories, in 0 collections
- Scope
- Repo-internal: written for the repository it lives in. Optimizes TensorRT-LLM's own PyExecutor backend internals
Judge review
Placed #3 of 6 in the judged category final for MLOps & model serving.
- Trigger
85 85 - Instructions
90 90 - Specificity
93 93 - Safety
80 80 - Disclosure
85 85 - Maintenance
85 85
A disciplined profile-optimize-validate loop with concrete env vars and CPU-affinity commands, cleanly cross-linked to a sibling bottleneck-confirmation skill.
Not checked yet
- No blind task eval yet
- Sibling files (scripts, references) not fetched; security scan covers SKILL.md only
Skillspace indexes and rates this skill; it does not host or endorse it. Run third-party skills at your own risk.