MCP App StoreMCP App Store
from ai-infra-auto-driven-skills699
Analyze LLM torch profiler traces to identify layer-level bottlenecks, kernel flows, and timing gaps in forward passes.
This skill provides high-resolution analysis of LLM execution pipelines by processing Chrome-trace JSON files from torch profiler. It transforms coarse trace data into actionable timing tables, enabling developers to pinpoint exactly which layers or kernels are causing performance degradation.
Activate this skill when a general profiler summary is too vague. It is especially useful for models with complex architectures (like DeepSeek-V4 NSA) where alternating layer types or specific anchor kernels are needed to delineate boundaries between attention and FFN blocks.
layer_timeline_analyzer.py for steady-state window detection, layer_kernel_breakdown.py for deep-dive compute flow analysis, and perfetto_time_mapper.py for UI navigation.Designed for agents with shell execution capabilities and filesystem access to model configs and trace files, such as Codex, Claude Code, or other ACP-enabled harnesses.
This skill has not been reviewed by our automated audit pipeline yet.
vLLM Qwen3 Coder Optimization
Guided, PR-backed manual for auditing, debugging, and extending the Qwen3 Coder tool parser in vLLM—focuses on schema edge cases, tool-parser regressions, and v
vLLM Qwen3.6 Optimization
Guidance and PR-driven validation steps for optimizing and documenting Qwen3.6 support in vLLM; use when auditing or implementing model-specific changes.
Artificial General Research (AGR)
An autonomous iterative optimization framework that uses a 'fresh-context-per-iteration' loop to optimize measurable code metrics.
SEO-AGI (SEOBuild Onpage)
A full workflow for generating SEO-optimised pages designed to rank and be cited by LLMs, using live SERP research, chunk architecture, and quality gates.
Graphsignal GPU Profiler
Set up and manage GPU profiling, tracing, and monitoring for AI inference workloads using vLLM, SGLang, and PyTorch.
MSBuild Evaluation Performance
Diagnose and optimize MSBuild project evaluation speed to reduce build startup times.
Karpathy Coding Guidelines
Implements Andrej Karpathy's behavioral guidelines to reduce LLM coding mistakes through simplicity and surgical changes.
Tencent Cloud PG Slow Query Diagnosis
Diagnoses slow SQL and performance anomalies for Tencent Cloud PostgreSQL using direct OpenAPI calls.