An OpenCode plugin that converts documents to Markdown using Microsoft MarkItDown.
- PDF — Extract text from academic papers, reports, slides
- DOCX / PPTX — Office document conversion
- HTML — Web page to Markdown
- Images — OCR for text in images
- EPUB, CSV, XLSX — Various file format support
- URL — Direct conversion from web URLs
- uv (>= 0.10) — Python package manager, automatically manages markitdown's environment
- OpenCode installed
# Linux / macOS
curl -LsSf https://astral.sh/uv/install.sh | sh
# Or via Homebrew
brew install uvopencode plugin opencode-markitdownOr add to your opencode.json:
{
"plugin": ["opencode-markitdown"]
}Then restart OpenCode. On first use, uvx automatically downloads and caches Python dependencies (~30s). Subsequent calls are instant.
| Tool | Description |
|---|---|
markitdown |
Extract text content from a document file or URL as Markdown |
Once installed, the markitdown tool is available to OpenCode:
"Read this PDF and summarize the key points" "Extract the text from this DOCX file" "Convert that webpage to markdown"
The plugin uses uvx to run Microsoft's MarkItDown CLI:
uvx --from "markitdown[all]@..." markitdown <file>
- Zero Python environment management —
uvxhandles everything - Dependencies cached in
~/.cache/uv/after first run - No conda, no pip, no venv needed
MIT