67 tagged GPU, measured the same way as everything else here.
Browse within: deep-learning 29cuda 25cursor 19autograd 18neural-network 18numpy 18tensor 18infrastructure 16NVIDIA 12Orchestration 11cloud-functions 11kubernetes 11inference 10llm-inference 10
Skill Claude CodeCodex
Read VENDOR.md for the exact integration baseline, namespace mapping and local device differences. This package is independently owned by SM100/103; keep the original Rubin tree and frontend unchanged.
Skill Claude CodeCodex
Use with the dstack skill for model-serving work when the image, serving command, resources, backend/fleet choice, or service behavior is not proven. Guides task-first prototyping on real hardware, choosing fleets/backends that can reuse idle instances and caches, checking vLLM/SGLang sources, and verifying the final…
Skill Claude CodeCodex
Drive the tty7 terminal workbench from the shell with the tty7 binary — list workspaces/tabs/panes, split a pane, send text or keystrokes into one, capture what is on a pane's screen, run a command in a real PTY and pass its exit code through, run a command in a shell pane you already have and get back its output and…
Xiangyue-Zhang/auto-deep-researcher-24x7
Skill Codex
Daily arXiv paper recommendations with automatic deduplication.
Skill Claude CodeCodex
Generate the repository-specific validation and benchmark adapters that kbench will orchestrate. Use once when harness/prepared.json is absent.
Skill Claude Code ✓ vendor
Install, operate, and tear down self-hosted NVIDIA Cloud Functions (NVCF) deployments with nvcf-cli. Use for control-plane or compute-plane install, status checks, cluster registration, function deploy/invoke, task create/list/cancel/delete, API keys, admin tokens, JWKS rotation, failed-install diagnosis, and…
Skill Claude Code
Fix CI failures and performance regressions for a Julia package PR by iterating - triage the latest CI results, fix one root cause, verify locally, push. Use when a PR's GitHub Actions or Buildkite CI is failing, or when CI jobs run slower than they do on the main branch. Verifies CPU and GPU compilation locally…
Skill Claude Code
Fix CI failures and performance regressions for a Julia package PR by iterating - triage the latest CI results, fix one root cause, verify locally, push. Use when a PR's GitHub Actions or Buildkite CI is failing, or when CI jobs run slower than they do on the main branch. Verifies CPU and GPU compilation locally…
runpod/runpod-plugins-official
Skill Claude Code needs its repo
Add or change a node in the Runpod concept graph (plugins/runpod/skills/runpod-usage/concepts/.yaml), including process nodes that explain how something works step by step. Use when asked to add a concept, add a rule to one, document how a Runpod flow works, or fix a node. Covers duplicate checks, choosing the kind…
Skill Claude CodeCodex needs its repo
Use when (re)indexing a Sweet Search project. Runs the full-profile indexer with GPU model prewarming (CoreML cascade on M3+, candle Metal on M1/M2, ORT CPU elsewhere), kills ORT CPU models during indexing to avoid memory contention, and rewarms them for query readiness on completion. Incremental runs under 20 files…
Skill Claude Code
Fix CI failures and performance regressions for a Julia package PR by iterating - triage the latest CI results, fix one root cause, verify locally, push. Use when a PR's GitHub Actions or Buildkite CI is failing, or when CI jobs run slower than they do on the main branch. Verifies CPU and GPU compilation locally…
KempnerInstitute/hpc-agentic-recipes
Skill Claude CodeCodex
Search the web, arxiv, Crossref, PubMed, OpenAlex, or Wikipedia, and fetch page text, using common/tools/search.sh. Use whenever the user asks to search online, look something up, or find papers while served by a local model, and whenever the built-in WebSearch tool fails or is silently dropped by the endpoint.
Skill Claude CodeCodex
Local speech-to-text using faster-whisper. 4-6x faster than OpenAI Whisper with identical accuracy; GPU acceleration enables 20x realtime transcription. SRT/VTT/TTML/CSV subtitles, speaker diarization, URL/YouTube input, batch processing with ETA, transcript search, chapter detection, per-file language map.
troycheng/cuda-kernel-optimizer
Skill Codex
Use when optimizing or diagnosing GPU workload performance with CUDA, CUTLASS, Triton, PyTorch, vLLM or TensorRT-LLM; interpreting NCU, Nsys or PyTorch Profiler evidence; or preparing the workload and correctness checks for such optimization. General coding, skill maintenance and conceptual GPU questions do not…
Skill Claude CodeCodex
Profile inside a CuTe DSL (cutlass.cute) kernel with IKET (In-Kernel Event Tracing) and the run-iket profiler. Use when developing or tuning a CuTe DSL kernel and you need per-warp timelines of kernel phases (setup, mainloop, epilogue), producer/consumer (TMA/MMA) activity, or time spent in pipeline and mbarrier…
Skill Claude CodeCodex
Rent a GPU on vast.ai, run experiments on it over SSH/rsync, pull artifacts back, and destroy it. Use when a task needs a GPU (training, CUDA, large-model inference).
Skill Claude CodeCodex
Use AutoDL safely through its official developer APIs and documentation. Trigger for AutoDL balance, Container Instance Pro, elastic deployment, images, GPU stock, NFS, duration packages, WeChat notifications, container metrics, billing, storage, data retention, SSH/JupyterLab/VSCode, CUDA environments…
ssmall256/mlx-metal-kernels-skill
Skill Claude Code
Provides guidance for writing custom Metal compute kernels using MLX's mx.fast.metalkernel() API for Apple Silicon GPUs (M1, M2, M3, M4). Covers kernel patterns for RMSNorm, LayerNorm, softmax, attention variants (causal, sliding window, GQA), and reductions. Includes profiling, debugging, simdgroupmatrix MMA (M3+)…
Skill Codex
Use Project Bourne to plan, execute, reproduce, inspect, or trace scientific and engineering workloads when durable provenance matters. Trigger for simulations, numerical solvers, ML or training, GPU and HPC runs on local compute, Slurm, PBS, or LSF, reproducible experiments, failed attempts, experiment comparisons…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: