DeepSeek Harness Plugins · Model Inference
Plugins that extend DeepSeek Harness (DSH) with vision, workflows, memory, web tools and more.
A DeepSeek Harness plugin that detects cigarette objects in images, outputting bounding boxes, confidence scores, and labels via URL, base64, or file path input.
An MCP-based M/M/1 queue simulation plugin offering config validation, theoretical metric calculation, SimPy discrete-event simulation, and result comparison analysis.
A pedestrian detection plugin for model inference that accepts image URLs, Base64 data, or file paths and returns bounding boxes, confidence scores, and labels.
A license plate recognition plugin powered by xby API, identifying plate numbers, colors, single/double-layer plates, and bounding boxes via image URLs, Base64 strings, or file paths.
An AI detection toolkit for DSH offering 20+ vision capabilities including object detection, image classification, insect/plant/animal recognition, human pose estimation, and instance segmentation.
OCR plugin for business licenses that extracts credit codes, names, legal reps, capital, dates, scopes, and addresses via URL, Base64, or file path input.
A text conversion MCP plugin for development workflows, covering case conversion, encoding/decoding, JSON/XML/SQL formatting, hashing, and regex operations.
Turn CSV, Excel, or DuckDB data into interactive charts with zoom, filter, group, and SQL support, handling millions of rows smoothly.
Detect all wild animals in an image, returning bounding boxes, confidence scores, and labels; supports input via URL, BASE64, or file path.
A DeepSeek Harness plugin that recognizes pet facial expressions from image URLs, BASE64 strings, or file paths and classifies them into four emotions: Angry, Happy, Relaxed, and Sad.
Extract images from local files, URLs, or base64 data and convert them for LLM visual analysis, supporting OCR, object recognition, and content understanding.
A DeepSeek Harness dish-recognition plugin that accepts image URLs, BASE64 data, or file paths and returns probable dish names with confidence scores.