OpenDocKit
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPENDOCKIT_MCP_ROOTS | No | Folders the server may read, separated like PATH. Without it, any file the server's user can read. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Converts local documents (PDF, DOCX, PPTX, XLSX, HWP/HWPX, ODT/ODS/ODP, RTF, EPUB, HTML, email, CSV, text) to Markdown, keeping tables, headings and lists. For a long document call document_outline first, then convert_document with sections to read the pages, slides or sheets you need.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| convert_documentA | Convert a local document to Markdown. path: the file to read (absolute, or relative to where the server was started). sections: which pages, slides or sheets to return, by their numbers in document_outline: "3" or "2-5". Leave out for the whole document. max_chars: the most characters of the document to return (default 100,000). A longer text is cut there, and a note after it says so. |
| document_outlineA | Describe a local document without returning its text: format, title, size, and its pages, slides or sheets with their numbers, titles and lengths. Use it to decide which path: the file to read (absolute, or relative to where the server was started). |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 2 tools
The two tools have clearly separate purposes: `document_outline` returns metadata and structural information, while `convert_document` returns extracted text. Their descriptions explicitly explain how they work together, leaving no realistic chance of confusion.
Both names use snake_case and are descriptive, but the pattern is not fully uniform: `convert_document` is verb-first while `document_outline` is noun-first. This is a minor deviation and remains readable and predictable.
The server is very tightly scoped with only two tools. Both earn their place, but the surface is thin for a document-handling server and leaves little room for adjacent operations.
The core workflow is covered: inspect a document's structure, then convert selected sections to Markdown. Minor gaps exist, such as discovering supported formats or handling batch operations, but agents can reasonably work around them.