@jiaoqsh/dsh-document
DeepSeek Harness bundle: the read_document tool reads Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV, and PDF files as Markdown for the model
AI 분석
核心用途是让 AI 模型能够阅读和解析各种办公及电子书文档。适合需要处理复杂非文本格式文件的用户。支持设置文件大小上限及 PDF 最大页数。
설치
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:jiaoqsh/dsh-documentConfiguration
The bundle's layer inserts one row, document-tools, with schema defaults. Override it by id in your profile's cordis.patch.yml; a patch replaces the whole config, so restate every key you need:
- id: document-tools
config:
maxInputBytes: 104857600 # 100 MiB
readLimit: 2000
maxLineLength: 2000
maxOutputBytes: 51200
pdfMaxPages: 100
pdfProfile: fidelity
| Key | Default | Meaning |
|---|---|---|
maxInputBytes | 52428800 (50 MiB) | Inclusive byte cap on the source file. Enforced by the filesystem provider before any bytes are buffered; larger files are refused. |
readLimit | 2000 | Default and maximum number of Markdown lines returned by one call. |
maxLineLength | 2000 | Maximum characters per returned line; overflow is cut with … [line truncated]. |
maxOutputBytes | 51200 (50 KiB) | Maximum bytes of line text returned by one call; the window stops early and the footer says how to continue. |
pdfMaxPages | 100 | Maximum distinct pages one pages selection may name. |
pdfProfile | fidelity | PDF Markdown profile: fidelity keeps source structure, compact spends fewer tokens. |
Every numeric value must be a positive integer and pdfProfile one of the two names; anything else fails the plugin load with a message naming the key.