# t0md
t0md converts PDF, Word, PowerPoint, HTML, and JSON files into clean Markdown for people and AI agents. It also provides a remote MCP server for converting documents inside compatible agent tools.
- Findings24:
- 網站:
- 價格: Freemium
- 分類: AI文件擷取, AI開發者工具, MCP 工具
## 什麼是 t0md？
t0md is a document-to-Markdown converter for PDFs, Microsoft Word files, PowerPoint presentations, HTML, and JSON. Upload a supported file in the browser to preview the result, copy the Markdown, or download a `.md` file. Browser conversion is free, requires no sign-up, and accepts files up to 25 MB.
The product is intended for workflows where documents need to be readable by people, searchable and diffable in text tools, or supplied to language models. It preserves common structures such as headings, lists, and paragraph breaks, while tables and complex layouts are handled on a best-effort basis because PDFs are layout-oriented rather than semantic.
t0md also provides a remote MCP server for compatible AI agents. The recommended streamable-HTTP endpoint is `https://t0md.com/mcp`; a legacy SSE endpoint is available for older clients. Agents can call `convert_to_markdown` to convert supported files during a conversation, receive Markdown inline, and get a temporary download link.
## t0md 能做什麼？
### Five live input formats
Converts PDF, Microsoft Word (\`.docx\`), PowerPoint (\`.pptx\`), HTML, and JSON into Markdown. EPUB, XLSX, image OCR, and URL conversion are listed as upcoming rather than available now.
### Browser preview and export
Upload a file in the web interface, inspect the Markdown preview, copy the result, or download it as a \`.md\` file. Browser uploads are limited to 25 MB.
### Structure-aware PDF extraction
PDF text is extracted with layout information, then processed with heuristics that identify lists, promote short all-caps lines to headings, collapse excess blank lines, and rejoin words split by hyphenation.
### Remote MCP conversion
The \`convert\_to\_markdown\` MCP tool lets compatible agents convert supported documents mid-conversation. It auto-detects format from the filename and content and returns Markdown inline with a download link valid for 10 minutes.
### HTTP API and staged uploads
A multipart \`POST /convert\` endpoint returns JSON containing Markdown, filename, page count, byte count, and conversion duration. \`POST /mcp/upload\` provides an upload ID for larger agent workflows.
### In-memory processing
The service states that uploaded documents are processed in memory and discarded. Generated MCP download links are retained for at most 10 minutes.
## 使用情境
### Prepare research papers for an AI assistant
Convert academic papers or reports into text-oriented Markdown that can be pasted into an LLM context or processed by an MCP-compatible agent.
### Create reusable notes from documents
Move converted output into tools such as Obsidian, Notion, Bear, or a GitHub README instead of relying on screenshots or manually repaired copy and paste.
### Review document revisions in Git
Use Markdown as a text-based representation for comparing drafts, tracking contract changes, and identifying paragraph-level revisions.
### Convert files during an agent session
Connect the MCP server to Claude Code, Claude Desktop, Cursor, Windsurf, OpenClaw, Hermes, VS Code with an MCP extension, or Continue.dev so an agent can request conversion without leaving the chat.
### Automate conversion through HTTP
Send supported files to \`/convert\` from an application or script and consume the returned JSON fields for Markdown, filename, pages, bytes, and processing duration.
## 常見問題
### How do I convert a document in the browser?
Drop a PDF, Word, PowerPoint, HTML, or JSON file into the converter or choose it from your device. The result appears as a Markdown preview that you can copy or download as a \`.md\` file. No account is required.
### Which formats are currently supported?
PDF, Microsoft Word, PowerPoint, HTML, and JSON are live. EPUB, XLSX, image OCR, and URL-to-Markdown conversion are marked as coming soon.
### How reliable is the output for complex PDFs?
Headings, bullet lists, numbered lists, and paragraph breaks are preserved where detected. Tables and complex layouts are best-effort because PDFs store visual layout rather than semantic document structure, so converted output should be reviewed when formatting is important.
### How does the MCP integration work?
Add \`https://t0md.com/mcp\` as a streamable-HTTP MCP server in a compatible client, then ask the agent to convert a document. The primary tool is \`convert\_to\_markdown\`; it returns Markdown inline and a one-time download URL. The SSE endpoint at \`https://t0md.com/mcp/sse\` is available for older clients.
### Are uploaded files stored, and is an API key required?
The site states that browser uploads are converted in memory and discarded. MCP-generated download links expire after 10 minutes. The MCP server is available without an API key today.
## 快速資訊
- **Product type**: Document-to-Markdown converter
- **Live formats**: PDF, Microsoft Word, PowerPoint, HTML, JSON
- **Browser limit**: 25 MB per file
- **Agent protocol**: Remote MCP server with streamable HTTP and legacy SSE
- **Browser access**: Free, no sign-up
- **Website**: t0md.com