Chensihakniroth

The registry covers LLM fine-tuning (LoRA, DPO, GRPO), distributed training (FSDP, DeepSpeed), inference serving (vLLM, TensorRT-LLM), vector search (Qdrant, Pinecone, FAISS, Chroma), financial Excel modeling, web scraping, OSINT investigation, penetration testing, image/video/music generation, and GitHub pull-request lifecycle management.

ML engineers training and deploying language models, investment analysts building valuation models, security researchers conducting authorized pentests and supply-chain forensics, data scientists curating training corpora with NeMo Curator, and creative technologists producing video, comics, and generative art.

Skills are loaded by the Anakot Agent runtime on Linux, macOS, or Windows. Most declare pip dependencies (torch, transformers, vllm) or CLI prerequisites (curl, jq, gh, ffmpeg) in their frontmatter; credential-based skills like Shopify, Notion, and Google Workspace prompt for API tokens during setup.

Nearly all skills carry MIT licenses, with the Anthropic-adapted finance models under Apache-2.0. The PowerPoint skill is proprietary with separate LICENSE.txt terms. External services such as Lambda Labs, Modal, Pinecone, and Hyperliquid may incur their own usage costs.

ML skills require PyTorch, Transformers, and often NVIDIA GPUs with CUDA; NeMo Curator needs RAPIDS. Security skills need CLIs like sherlock. Finance skills need openpyxl. Several skills require API keys (Shopify, Airtable, Notion, USDA) or desktop apps such as Blender 4.3+, Obsidian, or a Muse/OpenBCI headset for neuroskill-bci.