liger-kernel-perf

Optimizes the performance of existing Liger Kernel Triton kernels. Profiles kernels, diagnoses bottlenecks (memory-bound vs compute-bound), generates multiple optimization variants with benchmarking, and applies the best variant while maintaining correctness. Supports GPU architecture-specific optimization (Ampere, Hopper, Blackwell). Use when a user asks to optimize, speed up, tune, profile, or reduce memory of an existing Liger kernel.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

85%Weight 40%Scale 1-5

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

A well-structured, concise orchestration skill with strong workflow clarity and validation. Its main weakness is that the detailed reference files it delegates to are missing from the bundle, breaking the progressive-disclosure structure.

Suggestions

Bundle the referenced files (profiler.md, optimizer.md, finalizer.md, optimization-strategies.md) under references/ so the delegated workflows actually exist.

Add the templates/ directory with optimization-profile.md and variant-notes.md that the pipeline references, or inline minimal template skeletons into SKILL.md.

Verify each markdown link target resolves to a real bundled file; broken references undermine the otherwise clear navigation.

Dimension	Reasoning	Score
Conciseness	Lean body that uses tables and numbered lists, assumes Claude's competence (no explanations of Triton/GPU basics), and has every token earning its place.	5 / 5
Actionability	Provides concrete file paths and executable commands (pip install -e, pytest, make checkstyle, ruff) with explicit thresholds, but core optimization detail is delegated to referenced files that are not present in the bundle, leaving a minor gap.	4 / 5
Workflow Clarity	Clear 3-stage sequence with explicit human checkpoints, pre-flight validation, concrete stop conditions, and feedback loops (smoke-test discard, full-test hard gate, checkstyle auto-fix retry) plus a guardrails table.	5 / 5
Progressive Disclosure	The in-skill structure is well organized with clearly signaled one-level references, but the referenced bundle files (profiler.md, optimizer.md, finalizer.md, optimization-strategies.md, templates/) do not exist in the workspace, so the claimed progressive-disclosure structure is not actually backed by a bundle.	3 / 5
	Total	17 / 20 Passed

Description

92%Weight 40%Scale 1-5

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

A strong, specific description that clearly states what the skill does and when to use it with natural trigger phrases. Minor room to add synonyms to the trigger list for maximum keyword coverage.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions (profiles, diagnoses bottlenecks, generates variants with benchmarking, applies best variant, supports arch-specific optimization) with comprehensive coverage, matching the top anchor.	5 / 5
Completeness	Explicitly answers both what (the profiling/diagnosis/variant/benchmark/apply pipeline) and when (a concrete 'Use when a user asks to optimize, speed up, tune, profile, or reduce memory...' clause), matching the top anchor.	5 / 5
Trigger Term Quality	Includes natural trigger phrases users would say (optimize, speed up, tune, profile, reduce memory) but lacks some synonyms (accelerate, improve throughput, make faster), sitting just below the comprehensive-coverage anchor.	4 / 5
Distinctiveness Conflict Risk	The Liger Kernel Triton-specific framing with named GPU architectures (Ampere, Hopper, Blackwell) creates a clear niche with distinct triggers and minimal conflict risk.	5 / 5
	Total	19 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 10 missing	Warning

	Total	15 / 16 Passed

Repository: linkedin/Liger-Kernel
Path: .agents/skills/liger-kernel-perf/SKILL.md
Commit: 7777995

Reviewed: 4 days 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.