Doc2md

提供 doc2md 命令行工具，通过 doc2md 平台 API 把本地 PDF、DOCX、PPTX 转换为 Markdown。支持单文件、整目录递归与并行转换，可配置转换后端、超时和覆盖策略；需自备访问令牌并联网，且需注意文件会上传到该服务。

yance-dev

热度 16.02

最近更新 2026-06-11 11:47:48

分类：效率依赖：读写文件 / 第三方 API应用场景：文件转换 / 工作流

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add doc2mdyc

Metadata

name: doc2md
description: Use when the user wants to convert local PDF, DOCX, or PPTX files into Markdown with the packaged doc2md CLI, especially for batch conversion, recursive folder processing, or Windows and Linux command-line usage.

Doc2md

Overview

This skill provides a packaged doc2md CLI for converting local PDF, DOCX, and PPTX files into Markdown through the doc2md platform API.

Use this skill when the user needs a ready-to-run command-line workflow instead of writing custom conversion code.

Bundled Files

Windows binary: scripts/doc2md-cli.exe
Linux binary: scripts/doc2md-cli

Both binaries are statically linked and do not require an extra runtime.

Prerequisites

A valid doc2md bearer token
Network access to the doc2md API service
Input files in PDF, DOCX, or PPTX format

Configuration can be provided with either:

Environment variables: DOC2MD_API_BASE_URL, DOC2MD_BEARER_TOKEN
Config file: ~/.doc2md/config.json

Environment variables take precedence over the config file.

Usage

Windows PowerShell

$env:DOC2MD_BEARER_TOKEN = 'your-jwt'
$env:DOC2MD_API_BASE_URL = 'http://192.168.99.85:5173'

.\scripts\doc2md-cli.exe -output-dir .\converted .\document.pdf

Linux

export DOC2MD_BEARER_TOKEN='your-jwt'
export DOC2MD_API_BASE_URL='http://192.168.99.85:5173'

./scripts/doc2md-cli -output-dir ./converted ./document.pdf

Common Commands

# Convert one folder recursively
./scripts/doc2md-cli -output-dir ./converted ./docs/

# Convert multiple inputs in parallel
./scripts/doc2md-cli -output-dir ./converted -concurrency 4 ./doc1.pdf ./doc2.docx ./folder/

Flags

Flag	Default	Description
`-output-dir`	required	Directory that receives extracted results
`-converter`	`mineru`	Backend converter: `mineru` or `marker`
`-recursive`	`true`	Scan directories recursively
`-keep-zip`	`false`	Keep `result_clean.zip` after extraction
`-overwrite`	`true`	Overwrite existing output directories
`-concurrency`	`1`	Number of files processed in parallel
`-poll-interval`	`5s`	Job polling interval
`-job-timeout`	`24h`	Per-file timeout
`-http-timeout`	`2m`	Per-request HTTP timeout

Output Behavior

Each input document is written to its own subdirectory under -output-dir
The CLI prints timestamped progress and per-file status to stdout
Ctrl+C cancels remaining work gracefully
The process exits non-zero when any conversion fails

Config File Example

{
  "apiBaseUrl": "http://192.168.99.85:5173",
  "bearerToken": "your-jwt"
}

Save this file as ~/.doc2md/config.json when environment variables are not convenient.

Troubleshooting

Authentication failure: verify DOC2MD_BEARER_TOKEN
Connection failure: verify DOC2MD_API_BASE_URL and service reachability
Empty output directory: confirm the input file type is supported and the API job completed successfully
Existing output overwritten unexpectedly: pass a safer output directory or adjust -overwrite