dsh-qps-calc
Run the following command in DeepSeek Harness:
dsh plugin install uckkk/dsh-qps-calc
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install uckkk/dsh-qps-calc in DeepSeek Harness to install this plugin, source code: https://github.com/uckkk/dsh-qps-calc
About this plugin
When an inference service is about to launch or scale up, figuring out the true peak QPS often boils down to a gut feeling rather than a reproducible calculation. dsh-qps-calc was built to close that gap. It registers a single tool called qps_calc: you supply the key workload parameters, the plugin runs them through exact formulas, and returns a structured, detailed breakdown you can drop straight into a capacity review or a load-test plan.
The entire calculation is implemented in pure Node.js with zero network calls, so it runs reliably on an offline dev machine or in a CI pipeline, and your parameters never leave the local process. If you own capacity planning for an LLM inference service, need a theoretical ceiling reference before a stress test, or want the whole team aligned on a shared number during a design review, dsh-qps-calc acts as a lightweight desk calculator: validate assumptions in seconds and skip the hassle of rederiving formulas every round.
Use Cases
- Peak capacity assessment before launching an LLM inference service
- Establishing a theoretical QPS ceiling as a baseline before load testing
- Aligning the team on peak load figures during design reviews
- Comparing QPS shifts across different parameter combinations
Best For
- Capacity planning leads for LLM inference services
- Inference engineers who need to quickly validate peak load assumptions
- Technical leads responsible for load-test and capacity design
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.