dsh-docling
为 DeepSeek Harness 提供本地 PDF、Office 文档、图片及 OCR 的文档智能解析服务。
AI 分析
用于本地解析PDF、Office和图片并支持OCR。适合需要离线处理多格式文档、提取文本信息的DeepSeek Harness用户。需配置Python或Node运行环境及Tesseract语言包。
安裝
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:Sqhao-O/dsh-docs說明文件
閱讀完整 README ↗Configuration
| Field | Default | Meaning |
|---|---|---|
engine | auto | node, python, or auto; auto selects configured embedded Python, otherwise Node Xberg. |
runtimeDir | unset | Absolute embedded-runtime directory. |
pythonCommand | unset | Trusted Python executable for a managed runtime. |
pythonWorkerPath | shipped worker | Absolute Python worker override. |
tessdataPath | runtime ocr/tessdata | Absolute bundled Tesseract language-data directory. |
ocrBackend | auto | auto or tesseract; both select the pinned local Tesseract backend. |
ocrLanguages | [eng] | Ordered local OCR language packs. |
timeoutMs | 120000 | Per-conversion deadline. |
maxFileBytes | 52428800 | Authorized input-size cap. |
allowedLocalRoots | [] | Absolute non-root directories the model may read. |
defaultOcr | false | OCR default for images and scans; enable it only with a configured local tessdata runtime. |
defaultTableMode | accurate | fast or accurate PDF table behavior. |
defaultOutputFormat | md | md, text, or json. |
maxOutputChars | 32000 | Maximum result returned to the model. |
debug | false | Logs safe engine metadata only. |
Older baseUrl, apiKey, enableRemoteUrls, and allowPrivateUrls profile
fields are accepted only for migration; they do not enable a remote engine.