BD perf-optimization-casebook
Casebook of past successful and classic TensorRT-LLM optimizations (runtime/execution and kernel level) recorded as reusable decision precedents. Consult when deciding which optimization to apply for a classified bottleneck or a given config/model/hardware, to find prior art and adapt a proven approach instead of guessing. Each case records applicability signals, mechanism, how to apply, expected effect, accuracy risk, verification, and rollback.
Promises a TensorRT-LLM optimization casebook: ready-made solutions for runtime and kernel bottlenecks. Each case includes applicability signals, mechanism, instructions, and risks. In practice: 49 files with half the references broken. Text quality scores 76, process integrity 39. No critical findings, but structure limps—no model runs show whether recommendations actually save time or just add lookup overhead. Safety is clean (100), but missing execution data leaves accuracy impact unclear.
Install if you already know your bottleneck and need precedents to adapt. If you're hunting an unknown problem, the casebook may waste time with incomplete patterns.
Casebook of past successful and classic TensorRT-LLM optimizations (runtime/execution and kernel level) recorded as reusable decision precedents.
As a process D 39/100 · Unfinished process — References files that are not bundled: references/<family>/, references/<family>/<slug>.md
How to improve
- 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: 49. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- warning
missing-refreference to a missing file: references/<family>/ - warning
missing-refreference to a missing file: references/<family>/<slug>.md - note
edit-residuethe text marks something as outdated (lines 76): check that old rules are not kept next to new ones — the full check reads the text for contradictions
Process rating: all ten parameters 39/100
- 0Tools and files. 2 referenced file(s) missing: references/<family>/, references/<family>/<slug>.md
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Progress reporting. Says nothing while it works
- 20When it triggers. No condition that starts the skill
- 30Running it twice. 15 mutating operations with no state check
- 40Result and completion. Does not say what the result is
- 50Failures and branches. 0 branches, has a failure section
- 70Execution cost. Instruction body is 4500 tokens
- 85Steps. 43 steps, 2 vague phrases
- 100Consistency. Name and required fields are in place
- 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)
- +3Output format is not stated: the model decides each time
- +4No input/output examples
- +2Single-language instructions
- +3Description length 450: enough signal without eating the budget
- +4Structure: 12 headings
- +3Step-by-step instructions: 43 items
- +4Reference files are cited in the instructions (1 of 1)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 76.