CC perf-optimize
Launch and operate this repo's perf-optimize workflow, which iteratively APPLIES TensorRT-LLM serving optimizations — baseline benchmark at one concurrency or a Pareto curve of them (tok/s/user vs tok/s/gpu), analytical SOL projection on by default (via the internal-perf-sol-analysis skill) sizing the headroom the campaign chases, profile-ranked roadmap.yaml (nsys + ncu per-kernel analysis via the perf-nsight-compute-analysis skill), a fixed budget of rounds applying items one at a time gated on measured gain (curve mode uses a Pareto gate; the evaluator accepts, rejects, or pushes back each attempt and nsys-profiles every accept), one final-verification QA benchmark, expected-vs-measured report with Pareto improvement results. Use when the user wants to optimize / improve / speed up a trtllm-serve deployment (throughput, TTFT, TPOT, ITL, e2e latency) or says "run perf-optimize". For diagnosis WITHOUT applying changes, use the perf-analyze workflow instead.
The skill launches a TensorRT-LLM optimization workflow: baseline benchmark, Pareto curve (throughput per user vs per GPU), analytical performance ceiling, kernel profiling via nsys and ncu, iterative optimization with measured gain gates at each step. Promises to automate bottleneck discovery and validate results.
Catch: single folder instead of complete file set, code quality 66/100, process score 54/100. No scripts or concrete run examples. Never tested on actual models. Missing file references and unclear workflow dependencies mean you risk broken links and unmet requirements.
Don't install: too many missing pieces and no real-world runs to confirm it actually works.
Launch and operate this repo's perf-optimize workflow, which iteratively APPLIES TensorRT-LLM serving optimizations — baseline benchmark at one concurrency or…
As a process C 54/100 · Has gaps — References files that are not bundled: examples/disaggregated/slurm/benchmark/submit.py
How to improve
- The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
- The text references files that are not there: add them or drop the references.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 1. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
body-longSKILL.md body ≈ 5127 tokens (recommended < 5000); move details to references/ - warning
missing-refreference to a missing file: examples/disaggregated/slurm/benchmark/submit.py
Process rating: all ten parameters 54/100
- 0Tools and files. 1 referenced file(s) missing: examples/disaggregated/slurm/benchmark/submit.py
- 0Inputs and preconditions. Does not say what the process needs to start
- 30Running it twice. 9 mutating operations with no state check
- 40Result and completion. Does not say what the result is
- 70When it triggers. States when to use, but not when not to
- 70Execution cost. Instruction body is 5127 tokens
- 100Steps. 36 steps
- 100Failures and branches. 6 branches, has a failure section
- 100Consistency. Name and required fields are in place
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low The response is described with custom markup (9 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 971: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- +2Single-language instructions
- +4Structure: 8 headings
- +3Step-by-step instructions: 36 items
- +4Has examples (1 code blocks)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 66.