SJF-ECNU/dsh-cloakbrowser ↗★ 2
dsh-cloakbrowser
Native CloakBrowser tools for DeepSeek Harness.
AI 분석
核心用途是赋予智能体强大的网页自动化操作与视觉分析能力。适合需要执行复杂网页任务的用户。必要条件是需配置 Python 环境并安装 CloakBrowser 运行时。
설치
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:SJF-ECNU/dsh-cloakbrowser6. Complete first configuration
CloakBrowser runtime. The default commands install into ~/.dsh/venvs/cloakbrowser, which this plugin uses automatically. To confirm that environment is ready:
~/.dsh/venvs/cloakbrowser/bin/python -c "import cloakbrowser; print('CloakBrowser ready')"
If CloakBrowser is installed somewhere else, set the interpreter path in the same shell that starts DSH, then start or restart DSH:
export CLOAKBROWSER_DSH_PYTHON=/absolute/path/to/python
dsh web
Optional visual model. In the DSH web UI, open Settings → Plugins → Plugin configuration → Visual understanding model. Enter an OpenAI-compatible Base URL including its API version path (for example, one ending in /v1), an image-capable model name, select Chat Completions or Responses, enter the API key, and click Save. The API key is write-only: saving is the only time it is displayed or changed.
You can now ask the agent to open a page and describe its layout or images. browser_understand is called only when the agent needs visual grounding.