# 克隆技能到 skills 目录即可使用
# 首次运行自动安装所需依赖，开箱即用

# 清华源（推荐）
pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple

# 阿里源
pip config set global.index-url https://mirrors.aliyun.com/pypi/simple/

# 腾讯源
pip config set global.index-url https://mirrors.cloud.tencent.com/pypi/simple

# 仅本次生效（不修改全局配置）
pip install --index-url https://pypi.tuna.tsinghua.edu.cn/simple <package>

# PDF 转笔记
to-markdown pdf ./document.pdf

# 批量处理文件夹
to-markdown pdf ./pdf_folder/ --source folder

# 启用 OCR（繁体竖排）
to-markdown pdf ./ancient_book.pdf --ocr --ocr-dir vertical

# YouTube 视频转笔记
to-markdown video "https://youtube.com/watch?v=xxx"

# Bilibili 视频转笔记
to-markdown video "https://bilibili.com/video/BVxxx" --source bilibili

# 音频转笔记
to-markdown audio ./lecture.mp3

# 网页转笔记
to-markdown web "https://example.com/article"

# SOP 提取
to-markdown sop ./manual.pdf

# 强制重新处理
to-markdown pdf ./doc.pdf --force

# 指定输出目录
to-markdown pdf ./doc.pdf --output-dir ./my_notes

import asyncio
from src import ToMarkdown

async def main():
    tm = ToMarkdown(output_dir="./output")

    # PDF 转笔记
    result = await tm.pdf_to_note(
        path="./document.pdf",
        ocr_enabled=True,
        ocr_direction="auto",  # 自动判断横排/竖排
    )
    if result.is_success:
        print(f"输出: {result.data.output_path}")

    # 视频转笔记
    result = await tm.video_to_note(
        url="https://youtube.com/watch?v=xxx",
        source="youtube",
    )

    # 网页转笔记
    result = await tm.web_to_note(
        url="https://example.com/article",
        extract_main=True,
    )

    # SOP 提取
    result = await tm.extract_sop(path="./manual.pdf")

asyncio.run(main())

{
  "mcpServers": {
    "to-markdown": {
      "command": "python",
      "args": ["-m", "src.mcp.server"],
      "env": {
        "OUTPUT_DIR": "./notes",
        "CHECKPOINT_DIR": ".checkpoints"
      }
    }
  }
}

# skill.yaml
config:
  output_dir: "./output"
  checkpoint_dir: ".checkpoints"
  max_concurrency: 4
  retry:
    max_retries: 3
    base_delay_seconds: 1.0
    max_delay_seconds: 60.0
    timeout_seconds: 300.0
  ocr:
    language: "chi_tra+chi_sim+eng"
    dpi: 300
    direction: "auto"

IDLE → VALIDATED → RUNNING → CHUNKING → MERGING → COMPLETED
                         ↘     ↘          ↘
                          FAILED ←──────────┘
                            ↓ (retry < max)
                          RUNNING

.checkpoints/
├── pdf_to_note/
│   └── {input_sha256}.json    ← 每个任务独立 checkpoint
├── sop_extract/
├── video_to_note/
├── audio_to_note/
└── web_to_note/

from src.integrations import ObsidianIntegration

obsidian = ObsidianIntegration(
    vault_path="/path/to/obsidian/vault",
    default_subdir="notes",
)

# 保存笔记
obsidian.save_note(note, tags=["learning", "ai"])

# 创建索引页 (MOC)
obsidian.create_index(notes, index_name="学习笔记索引")

Result.success(data)        # 完全成功
Result.failure(error)       # 完全失败
Result.partial(data, errors) # 部分成功

"帮我把这个 PDF 转成 Markdown 笔记"
"把 /path/to/report.pdf 转成结构化笔记，启用 OCR"
"分析这个 SOP 文档，提取操作流程"
"总结这个 YouTube 视频的内容 https://youtube.com/watch?v=xxx"
"把这篇网页转成笔记 https://example.com/article"
"把 /path/to/audio.mp3 转录并总结成笔记"
"批量处理 /path/to/pdfs/ 文件夹下所有 PDF"

# Agent 内部通过 Python API 调用
from src import get_api
tm = get_api()(output_dir="./notes")

# PDF 转笔记
result = await tm.pdf_to_note(path="./doc.pdf", ocr_enabled=True)

# 视频转笔记
result = await tm.video_to_note(url="https://youtube.com/watch?v=xxx")

# 网页转笔记
result = await tm.web_to_note(url="https://example.com/article")

# 检查结果
if result.is_success:
    print(f"输出: {result.data.output_path}")