DeepSeek Harness PluginManifest Valid

picturereader

DeepSeek Harness image-understanding plugin for text-only models, with local image tools, OCR, document rendering, and an optional external VLM bridge.

VisionUI & ProductivityDeveloper ToolsSecurity & PolicySkills & Workflows

30GitHub Stars4ForksUpdated2026-08-20

Source

Install

$ dsh plugin --profile web add picturereader # npm 包；或从源码: dsh plugin --profile web add .

Plugin Overview & Capabilities

AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.

source-grounded

picturereader is an installable DSH plugin that adds image and document reading workflows to text-only models. It provides a visual-twin adapter for native image thumbnails, local pixel/OCR tools, configurable privacy-aware routing, document-to-image conversion, and optional OpenAI-compatible vision API calls.

Key Capabilities

Installs through `dsh plugin --profile web add picturereader` or headless-profile installation.
Adds image tools including scan, OCR, sampling, crop, palette extraction, comparison, batch processing, and unified vision analysis.
Provides a visual-twin adapter that exposes selected text models as image-capable `(视觉)` variants in DSH.
Supports privacy, smart, and strict routing modes for local evidence gathering and optional external VLM use.
Offers three OCR engine choices: built-in Windows OCR plus optional PaddleOCR and RapidOCR.
Converts PDF, Word, Excel, and PowerPoint documents into PNG pages using optional local dependencies.
Adds a DSH Web settings card for image-reading, OCR, routing, external VLM, and visual-twin configuration.

Useful For

Reading screenshots, charts, interfaces, and image-based text with text-only DSH models.
Running local OCR and pixel-level inspection for sensitive or offline image workflows.
Comparing images or regions, extracting palettes, and processing batches of images.
Rendering supported office documents or PDFs into images for page-by-page analysis.
Using an external OpenAI-compatible VLM for semantic image interpretation when configured.

Who It Fits

DeepSeek Harness users working with text-only models.
Users who need local image inspection and OCR tools.
Users who want optional, configurable external vision-model integration.
Users analyzing image-based documents in DSH.

Documented Limitations

WebP is not supported by image_scan and vision_analyze; images must be converted to PNG or JPEG first.
Native thumbnails for text models require enabling visual-twin models and restarting DSH.
Changes to selected visual bridge models require a DSH restart.
Optional PaddleOCR, RapidOCR, and document conversion require separate setup; document conversion also requires LibreOffice and PyMuPDF.
External VLM analysis requires a configured endpoint and may be unavailable offline.
The README notes that DSH attachment uploads default to about 5 MB, which may block large images before plugin processing.
The README recommends disabling `dsh-file-drop` because its image injection can conflict with the visual-twin/image bridge flow.

DSH Compatibility

Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.

Not tested yetNo runtime compatibility tests have been published yet.

Security Signals

Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.

Dependency Counts

package.json declares 3 runtime, 0 development, 3 peer, and 0 optional dependencies.

info

Dsh Bundle Declared

package.json declares DSH bundle metadata.

info

License Declared

GitHub reports the repository license as MIT.

info

Package Manifest Available

A root package.json was captured and can be inspected by the registry.

info

Source Available

Public GitHub source metadata is available for this registry snapshot.

info

Source & Registry Notes

Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.

Source repository: jing-hy/picturereader
Registry source: GitHub · dsh-plugin topic
Registry classification: Plugin
Source checked: 8cd65e1 · 2026-08-21

This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.