Loading...
Loading...
Found 41 Skills
Manage and monitor VSS alerts after the alerts profile is deployed. The deployment's mode (CV vs VLM real-time) is fixed at deploy time and determines the workflow — start/stop real-time alerts via the VSS Agent on a VLM deployment, onboard CV alerts by adding RTSP streams to VIOS on a CV deployment, query incidents, customize verifier prompts. Use when asked to start/stop a real-time alert, check or list alerts, add a camera, use a sample video for alerts, customize alert prompts, or view verdicts.
Use this skill when producing a VSS analysis report — Mode A per-clip VLM, Mode B incident-range via video-analytics. Not for real-time alerts or ad-hoc Q&A.
Use this skill when working with the RTVI VLM or RT-VLM microservice API on VSS 3.1. Generate dense captions and alerts for stored video files and live RTSP streams via `/v1/generate_captions_alerts`; upload media via `/v1/files`; add and remove live streams with `/v1/streams/add` and `/v1/streams/delete/{stream_id}`; call OpenAI-compatible `/v1/chat/completions`; consume Kafka caption, incident, and error topics; or debug rtvi-vlm responses. For deployment, read `references/deploy-rt-vlm-service.md` first.
Summarize a video by calling the VLM NIM or the Long Video Summarization (LVS) microservice directly. For short videos (under 60s) call the VLM's OpenAI-compatible chat completions endpoint; for long videos (60s or longer) call the LVS microservice. Use when asked to summarize a video, describe what happens in a video, analyze a recording, call or debug LVS summarize/model/health/recommended-config/metrics endpoints, or configure and troubleshoot the LVS service that backs long-video summarization.
Create basic Transformers model adapters for msModelSlim. Implements required interfaces and completes a four-step verification workflow: generate test model -> full fallback quantization -> weight verification -> quantization description validation. Use this skill when the user wants to: (1) create msModelSlim adapters for decoder-only LLM, (2) adapt understanding VLM text backbones for quantization, (3) implement W8A8/W4A16 quantization workflow for new models. Trigger: user mentions "msModelSlim", "adapter", "model adapter","quantization", "W8A8","W4A16", "transformers", "LLM", "VLM", "adapter creation", "适配器","模型适配", "量化", "模型适配器", "LLM量化"
Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.
Extract false-positive and false-negative gaps from VLM binary-classification-question (BCQ, yes/no) predictions. Use after running VLM evaluation when you have a predictions JSON and need to identify failure cases for DEFT root cause analysis on a binary-classification VLM workflow.
Multi-step video annotation pipeline that turns raw videos into Chain-of-Thought training data — multi-level captions, structured descriptions, and QA pairs (MCQ, binary, open-ended) with reasoning traces, via VLM/LLM distillation. Use when the user wants to "create video training data", "generate video QA datasets", "build CoT reasoning traces from videos", "auto-label videos", or run the video_reasoning_annotation pipeline. Triggers include "video annotation", "video CoT", "video QA", "chain-of-thought", "video captioning pipeline", "video distillation".
Build a slide deck whose source is text and whose figures are laid out by TALA — one Markdown file with a ```d2 fence per figure, compiled to a self-contained HTML deck (keyboard nav, overview grid, print-to-PDF), each figure as its own SVG, and a copy manifest. The deck is a page, so vlmkit's own gates read it: `check integrity` for a slide that breaks or clips, `check copy --manifest` for text the frame cut off, `check a11y contrast` for a projector. Loop: write `deck.md` → build → gates → fix. Use when asked for slides, a deck, a talk, a presentation, a review walkthrough, or a figure-heavy explainer that has to be presented rather than read. Not for one diagram (`d2-diagram`) or a moving figure (`explanatory-animation`).
Integrate with HyperAPI for financial document processing - OCR text extraction, document classification, PDF splitting, and structured data extraction from invoices, receipts, and financial documents. Use when the user needs to parse PDFs, extract text from documents, classify document types, split multi-document PDFs, or extract structured entities like invoice numbers, vendor names, line items. Keywords: hyperapi, hyperbots, document parsing, OCR, PDF processing, invoice extraction, receipt processing, document classification, VLM, vision language model.
Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.
Fine-tune any HuggingFace CV / VLM / LLM model on local NVIDIA GPUs inside an NGC PyTorch container. Use when the user wants to fine-tune a HuggingFace model (full or LoRA), train a vision / VLM / LLM model end-to-end, generate a reproducible HF training pipeline, smoke-test a HuggingFace model locally before scale-up, push a fine-tuned model to the HF Hub with a model card, or emit a self-contained rerun skill for an existing HuggingFace finetune. Supports image classification, object detection, semantic / instance / panoptic segmentation, depth estimation, image-text-to-text VLM (SFT / LoRA), and LLM SFT / DPO / GRPO. Six-step workflow: inspect and qualify, hardware and NGC image, research, generate and smoke, train + eval + infer, push and emit rerun skill.