AI Agent Hub
Back to plugins
🤖

xby-captcha

Model Inference Updated 2026.09.02

Run the following command in DeepSeek Harness:

dsh plugin install xby-skill/xby-captcha

Paste the following prompt into your AI chat to install this plugin:

Open a terminal in DeepSeek Harness and run dsh plugin install xby-skill/xby-captcha to install this CAPTCHA recognition plugin; source code is available at https://github.com/xby-skill/xby-captcha .

About this plugin

CAPTCHAs are often the last and most stubborn obstacle in automation scripts, RPA pipelines, and AI agent workflows. xby-captcha plugs into DeepSeek Harness and turns captcha recognition into a set of ready-to-call tools, so you can get structured answers from an image in a single conversation turn.

The toolkit covers four major captcha types. Text captchas handle digits, letters, and Chinese characters with confidence scores. Slider captchas offer gap detection via target-background matching, shadow-region diffing, and background-only gap location. Rotation captchas compute single-image angles (0-359 degrees) or relative angles between inner and outer rings. Click-to-select captchas detect icons or words in the order shown in a hint image, returning ordered coordinates. Bonus utilities include image similarity scoring and math expression solving (add, subtract, multiply, divide). Every tool accepts the input as a URL, a Base64 string, or a local file path, so it slots into any data pipeline.

This plugin is built for developers, automation engineers, and AI agent builders who need captcha recognition without spinning up their own model infrastructure. All calls are authenticated via an API key and executed in the DSH chat. Please respect the terms of service and applicable laws of every target site, and use these tools only where permitted.

Use Cases

  • Automation scripts that need to read text captchas (digits, letters, CJK)
  • RPA pipelines that must locate slider gaps or diff shadow regions
  • AI agent tasks requiring rotation-angle detection or ordered click-to-select recognition

Best For

  • RPA and automation engineers
  • AI agent builders
  • Web testers and scraping developers