graphsignal-python
GPU inference profiling, tracing, and performance monitoring
All Skills in This Repository (1)
Pure Emerald Level IndicatorsFrequently Asked Questions
FAQPage SchemaHow to install Graphsignal Profiler?▼
Run `npx skills add graphsignal/graphsignal-python --all -g -y` in your terminal to install this skill globally.
How to profile a vLLM server?▼
Set your GRAPHSIGNAL_API_KEY environment variable, then wrap your launch command with `graphsignal-run vllm serve <model> --port 8000` to start profiling automatically.
Does Graphsignal work with SGLang and PyTorch?▼
Yes. It auto-instruments SGLang, PyTorch, TensorRT-LLM, and dstack services, capturing Prometheus metrics and GPU profiling out of the box.
Will profiling slow down my inference workload?▼
No. The profiler runs as a separate sidecar process and uses low-overhead CUPTI APIs, so production performance impact is minimal.
Is my prompt data sent to Graphsignal?▼
No. The profiler never records prompts or completions and only makes outbound connections to api.graphsignal.com.
Related Repositories in Software Engineering
View All in Software Engineering→
openclawopenclaw
Run a personal AI assistant across your devices and chat apps
obrasuperpowers
Gives coding agents a disciplined workflow from idea to merged code
facebookreact
AI agent skills for building, testing, and porting React core