jiekesu967/dsh-markitdown0

dsh-markitdown

Microsoft MarkItDown as a DeepSeek Harness tool: convert PDF, Word, Excel, PowerPoint, HTML, CSV, EPUB, or a URL into Markdown the model can read.

AI Analysis

核心用途是为 AI 代理提供通用的文档转 Markdown 工具。适合需要让 AI 自动读取和解析各种格式本地文件或网页链接的任务。支持多种转换引擎配置。

Package
dsh-markitdown
Version
0.1.0
License
MIT
Last updated
Sep 10, 2026

Install

$npx -p @deepseek-ai/dsh dsh plugin --profile web add github:jiekesu967/dsh-markitdown

Configuration

Every field is optional; defaults are shown.

- id: markitdown
  name: dsh-markitdown
  config:
    engine: auto              # auto | markitdown | uvx | python | builtin
    uvxPackage: "markitdown[all]"   # slim it to "markitdown[pdf,docx,pptx,xlsx]"
    command: ""               # explicit markitdown executable, overrides PATH
    timeoutMs: 120000         # per-conversion deadline
    maxChars: 120000          # inline result cap; longer output is truncated
    maxBytes: 67108864        # bytes the built-in engine will read
    allowUrls: true           # accept http(s) inputs
    extraArgs: []             # extra CLI arguments for the external engine

engine: auto probes the chain once per plugin instance and caches the winner. Setting an engine explicitly disables the fallback: if you ask for uvx and there is no uv, the call fails with the reason and the fix, rather than quietly converting with something else.

A first uvx run downloads MarkItDown and its dependencies (about a minute). Package-manager progress lines are filtered out of the tool result; only real diagnostics are reported.