Optimize for GPU (Claude Skill)

GPU-accelerate Python code using CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuxfilter, cuVS, cuSpatial, and RAFT.

   
Type Claude Skill
Supplier K-Dense Inc. (community OSS)
Availability GA — part of the actively maintained K-Dense scientific-agent-skills collection
Pricing Free / OSS — license not stated upstream
Capabilities Read/Write — Claude runs the skill’s Python locally (Bash), not as an MCP tool

How to install

  • Claude Code / Claude.ai — Skills CLI (recommended): 
    npx skills add K-Dense-AI/scientific-agent-skills

    Installs the K-Dense collection; enable the optimize-for-gpu skill when prompted. Works across Claude Code, Cursor, and Codex via the Agent Skills spec (requires Node ≥ 18).

  • Claude Code / Claude Desktop — manual clone: 
    git clone https://github.com/K-Dense-AI/scientific-agent-skills
cp -r scientific-agent-skills/skills/optimize-for-gpu ~/.claude/skills/

    Project-scoped alternative: copy into .claude/skills/ instead of ~/.claude/skills/. The skill declares its own Python dependencies in its SKILL.md; install them (the K-Dense skills generally use uv / pip) when prompted on first use.

What it does

GPU-accelerate Python code using CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuxfilter, cuVS, cuSpatial, and RAFT. Use whenever the user mentions GPU/CUDA/NVIDIA acceleration, or wants to speed up NumPy, pandas, scikit-learn, scikit-image, NetworkX, GeoPandas, or Faiss workloads. Covers physics simulation, differentiable rendering, mesh ray casting, particle systems (DEM/SPH/fluids), vector/similarity search, GPUDirect Storage file IO, interactive dashboards, geospatial analysis, medical imaging, and sparse eigensolvers. Also use when you see CPU-bound Python code (loops, large arrays, ML pipelines, graph analytics, image processing) that would benefit from GPU acceleration, even if not explicitly requested.

Primary use cases: GPU-accelerate Python code using CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuxfilter, cuVS, cuSpatial, and RAFT.

Notes

Distributed as a SKILL.md (plus code examples) in the K-Dense collection — Claude executes it locally via Bash/Python rather than as an MCP server. Upstream license: not stated upstream. The skill name to enable after install is optimize-for-gpu.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.