Datathings Marketplace
datathings-marketplace is a Claude Code plugin marketplace for the Build phase that offers twelve plugins for GPU numerics, scientific libraries, and local LLM inference integration.
by datathings · github.com/datathings/marketplace
Register a 12-plugin marketplace when you need GPU numerics, local LLM inference, and power-system modeling helpers inside Claude Code driven builds.
Add it to Claude Code
Install the plugin in Claude Code. One command, paste-ready.
/plugin install datathings-marketplace@datathings/marketplaceBuilt to be called by your agent
Skillselion is itself an MCP server. Your agent can pull this entry and a paste-ready install config straight from the API - no copy-paste.
Retrieve this entry with skillselion.get_details("plugin:datathings/marketplace") and the paste-ready config with skillselion.get_install_config("plugin:datathings/marketplace").
What it does
datathings-marketplace is a Claude Code plugin marketplace with twelve plugins aimed at builders who live at the intersection of numerical computing, GPU acceleration, and optional on-device inference. Rather than generic productivity shortcuts, the keyword surface reads like a technical catalog: BLAS and LAPACK families, cuBLAS and cuFFT, graph-based kernels, embeddings pipelines, and references to greycat-style structured systems alongside power-network and short-circuit analysis tooling. Solo founders prototyping ML features, scientific SaaS, or energy-adjacent analytics can use it to keep heavy dependencies and API patterns one marketplace search away instead of re-explaining CUDA ports or inference server setup every sprint. The repo positions itself as community-sourced with a small star count, so treat each plugin as an integration accelerator you validate against your stack—Windows, Linux, ROCm vs CUDA, and offline vs server inference all change what installs cleanly. This is ideal when your agent must generate correct glue code around tensors, solvers, or local model runners, not when you only need marketing copy or SEO plugins.
Highlights
- Marketplace lists 12 plugins spanning GPU linear algebra, FFT, and sparse solvers
- Covers local and OpenAI-compatible inference paths including Ollama, llama.cpp, and vLLM
- Includes power-grid and pandapower-oriented modeling keywords for engineering domains
- Documents cross-platform paths: NVIDIA CUDA and ROCm (rocBLAS, rocFFT, etc.)
- Emphasizes batch computation, profiling, and multibackend numerical workflows
Why builders use it
Builders waste agent context re-deriving CUDA, BLAS, and inference-server setup every time they touch numerical or ML backend code.
After registration, you can enable domain-specific plugins so Claude Code follows vetted patterns for solvers, tensors, and Ollama or vLLM style inference in your repo.
At a glance
- Type - Plugin in Development Tools.
- Adoption - 0 installs, 8 stars, 0 votes.
FAQ
Who is datathings-marketplace for?
It is for developers and indie technical founders who combine scientific computing, GPU libraries, or local LLM serving with Claude Code assisted implementation.
When should I use datathings-marketplace?
Use it during backend or ML integration work when your agent must respect BLAS, CUDA, ROCm, or inference-server conventions instead of inventing APIs from scratch.
How do I add datathings-marketplace to my agent?
Add the datathings/marketplace entry to Claude Code, install the marketplace bundle, then turn on only the plugins that match your hardware and Python stack.
Comments
Share how you use datathings-marketplace, gotchas, or tips for other indie builders.
No comments yet - be the first to share how you use it.