deepseekplugins

DeepSeek Harness plugin

dsh-macos-vision-ocr

Offline macOS Vision OCR tool for DeepSeek Harness

Tools & Capabilities
Jump to install

Source facts

Repository
leozou320-ai/dsh-macos-vision-ocr
Latest update
Aug 16, 2026
Category
Tools & Capabilities
GitHub stars
0
Format
plugin
Catalog evidence
Upstream dsh.bundle evidence
Evidence path
package.json#dsh.bundle
Checked against
0.1.0-rc.8
Upstream check date
2026-08-20

This evidence comes from the upstream catalog. This site has not installed, run, or security-reviewed the plugin.

Install

Start with a prompt that asks an agent to review the GitHub repository and source. Switch to the command if you want to install it yourself.

Copy this prompt into DSH, Codex, or another agent and ask it to review the GitHub repository and source first.

Do not install or run any commands yet. Read this plugin's GitHub repository, README, and relevant source code. Then answer the questions below clearly and directly so I can decide whether it fits my needs:

1. What is this plugin, and what problem does it solve?
2. Who is it for, and what are its typical use cases?
3. How is it used after installation? Include one minimal example.
4. What known limitations or privacy, security, compatibility, or maintenance risks does it have?
5. Give a clear recommendation: recommend, conditionally recommend, or do not recommend, with reasons.

Distinguish statements documented by the repository, inferences from source code, and unknowns. If evidence is insufficient, say so explicitly. Do not guess or simply repeat the README.

GitHub: https://github.com/leozou320-ai/dsh-macos-vision-ocr
Plugin: dsh-macos-vision-ocr
Author: leozou320-ai

Check the source files

Read the README and other files from this plugin directory before installing.

File explorer4 files
README.mdSource · read only
README language

dsh-macos-vision-ocr

English | 简体中文

Offline OCR for DeepSeek Harness, powered by Apple's macOS Vision framework. The plugin adds an ocr_image tool that lets any text model extract text from screenshots, scans, and document images without an API key or network request.

Features

  • Runs locally with VNRecognizeTextRequest at accurate recognition level.
  • Supports PNG, JPEG, WebP, GIF, TIFF, BMP, HEIC, and HEIF.
  • Accepts BCP-47 recognition languages per call.
  • Compiles its small embedded Swift helper on first use, then reuses a content-addressed cache.
  • Returns bounded output and reports whether text was truncated.
  • Uses fixed subprocess argument vectors; image paths are never interpolated into a shell command.

Requirements

  • macOS 13 or later.
  • Xcode Command Line Tools with swiftc available on PATH.
  • DeepSeek Harness 0.1.0-rc.5 or a compatible developer-preview build.

This plugin intentionally fails on Linux and Windows because Apple Vision is not available there.

Install

Install directly from GitHub into any profile that should expose OCR:

dsh plugin --profile web add github:leozou320-ai/dsh-macos-vision-ocr

Restart the profile after installation. To remove it:

dsh plugin --profile web remove dsh-macos-vision-ocr

Usage

Ask the agent to read an image, or call the tool explicitly:

{
  "file_path": "./scan.png",
  "languages": ["zh-Hans", "en-US"]
}

The result contains the canonical image path, recognized text, selected languages, and a truncated flag.

Configuration

Edit this package's row in a later Harness patch layer when you need different defaults:

- id: dsh-macos-vision-ocr
  config:
    cacheDir: /absolute/path/to/cache
    languages: [en-US]
    maxOutputBytes: 2000000
KeyDefaultDescription
cacheDir$DSH_HOME/cache/ocrSwift source and compiled helper cache.
languageszh-Hans, zh-Hant, en-USLanguages used when a tool call omits them.
maxOutputBytes1000000Maximum captured OCR stdout per call.

Permissions, privacy, and security

  • OCR is local and the plugin makes no network requests.
  • Image paths are checked through Harness's filesystem service before the native helper runs, so the active filesystem policy still controls access.
  • The plugin runs swiftc once and then executes the cached native helper through Harness's subprocess service.
  • Recognized text becomes tool output and therefore enters the current session transcript and model context. Do not OCR material you would not send to the configured model provider.
  • Review third-party plugin source before installation and pin a commit for sensitive deployments.

Known limitations

  • Text recognition is not general visual understanding; it does not identify objects, faces, or scenes.
  • Reading order is a geometric approximation and can be imperfect for multi-column or highly stylized layouts.
  • The first call is slower because the Swift helper must compile.
  • Handwriting quality depends on language, image quality, and the macOS Vision version.

Development

node --check host.mjs
node --test
npm pack --dry-run

For a local install test:

dsh plugin --profile web add ./path/to/dsh-macos-vision-ocr

License

[MIT](LICENSE)

Back to searchOpen GitHub repository

More related plugins

View category