F
licenseA
qualityD
maintenanceEnables reading and extracting content from PDF documents including text (as Markdown), images, tables, and metadata from both local files and URLs, with OCR support for scanned documents.
2
-
doc2images-mcp
by BruceKangCN
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@doc2images-mcprender pages 1-3 of report.pdf at 200 dpi"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
doc2images-mcp
An MCP server that renders documents to images and returns them directly as multimodal content for the model.
Deterministic "document -> image" conversion only;
Does not convert documents to Markdown and does not do LLM-based document understanding / OCR;
PDF rendering is handled by pypdfium2 (PDFium), a self-contained Python wheel with no external binaries. PDFium ships its own encoding data, including the CJK CMaps, so Chinese/Japanese/Korean text renders correctly.
This version supports PDF only. The DOCX / PPTX branch is reserved in the code and will be enabled once LibreOffice is wired in.
Requirements
Dependency | Notes |
Python |
|
Dependency management and run entry point |
No system packages are required: pypdfium2 bundles the PDFium binaries.
Related MCP server: MokuPDF
Install
uv tool install doc2images-mcpWiring it into OpenCode
Add the following to the mcp.servers section of opencode.jsonc:
"doc2images": {
"type": "local",
"command": [
"uvx",
"doc2images-mcp",
],
},To cap how much image data reaches the context, you can also tighten OpenCode's image settings:
"media": {
"image": {
"auto_resize": true,
"max_width": 2000,
"max_height": 2000,
"max_base64_bytes": 5242880,
},
},Tool: pdf_to_images
Renders PDF pages and returns a list of images (one per page).
Parameter | Default | Description |
| — | Document path. Relative paths resolve against |
|
| Rasterization resolution, |
|
| First page to render (1-based) |
|
| Last page to render; |
|
| Hard cap on the number of pages (1–50) to protect the context |
|
| Maximum output width; wider pages are scaled down proportionally |
Environment variables
Variable | Default | Description |
| Current working directory | Base directory used to resolve relative paths |
Testing
Sample PDFs live in data/; the tests render them through the tool:
uv run pytest -vRendered intermediate images are written to tmp/, which is ignored via its
own .gitignore.
Roadmap
Office support: after installing LibreOffice, add
soffice --headless --convert-to pdffor.docx/.pptx/.pptin the dispatch function, then reuse the same rendering logic.Embedded image extraction: extract only the images embedded in a document instead of whole pages.
Alternative backend: PyMuPDF (
fitz) is a drop-in alternative if its richer text/embedded-image APIs are ever needed; the tool interface stays the same.
Available Tools
1 toolpdf_to_imagesPdf To ImagesA
Render document pages to images for vision models.
Returns one image per page. Use first_page/last_page/max_pages to limit how much of the document is sent to the model.
ParametersJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| dpi | No | Rasterisation resolution, 72-300. Higher is sharper but larger. | |
| file_path | Yes | Path to the document. Relative paths resolve against DOC2IMG_BASE_DIR (defaults to the current working directory). | |
| last_page | No | 1-based last page to render; 0 means "until max_pages". | |
| max_pages | No | Hard cap on the number of pages returned (1-50). | |
| max_width | No | Maximum output width in pixels; wider pages are scaled down. | |
| first_page | No | 1-based first page to render. |
TDQS
A4/5.0
Behavior4/5
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Without annotations, the description carries the behavioral burden, and it does disclose return cardinality ('one image per page') and that the volume sent to a model can be bounded. It doesn't cover resource cost, failure behavior on unsupported or corrupt files, or default page cap, so it is incomplete but substantive.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Conciseness5/5
Is the description appropriately sized, front-loaded, and free of redundancy?
Two short sentences, front-loaded with the verb and the output, and the second sentence routes straight to the relevant parameters. Nothing redundant.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Completeness4/5
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a six-parameter tool with no annotations and no output schema, the definition is close to sufficient: format, per-page cardinality, and page-limiting controls are all covered. It stops short of mentioning default output volume or resolution implications, and with no annotations the safety/cost profile is left unstated.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Parameters3/5
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the schema already names and constrains every parameter (dpi range, page bounds, max_pages cap, max_width). The description only gestures at first_page/last_page/max_pages as limiting controls, which is baseline-level added value at this coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Purpose5/5
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb and resource (render document pages to images) and frames the purpose around vision models, which tells an agent exactly when this output format is appropriate. No siblings exist, so differentiation isn't required.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Usage Guidelines3/5
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description says what the images are for ('for vision models') and points at three parameters for limiting volume, which implies usage context. However, it gives no when-not-to-use guidance or scope limits (e.g., at what point to prefer text extraction instead), so this is implied rather than explicit guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
1 tool update
v0.1.0- First observed
pdf_to_images
TDQS
A4.1/5.0
Scored across 1 tool
Disambiguation5/5
Only one tool is exposed, so there is no possibility of selecting the wrong tool. Its purpose—rendering PDF/document pages to images—is unambiguous.
Naming Consistency5/5
The sole tool uses clear snake_case (`pdf_to_images`), and there are no competing conventions to create inconsistency. Though not a strict verb_noun pattern, the name is readable and predictable.
Tool Count3/5
A single tool is thin for any server, but the server's scope is narrowly conversion. It earns its place, but 1 tool is borderline on the count rubric.
Completeness4/5
The tool covers the core PDF-page-to-image workflow, including page-range limiting. However, the server name suggests broader document-to-image support, while the only tool handles PDFs; this is a minor gap.
Maintenance
ActivityMaintained
ResponsivenessNo issues
Related MCP Connectors
Generate and read PDFs for AI agents: a generate_pdf and a read_pdf tool, priced per document.
Image & PDF tools for AI agents: compress, convert, resize, PDF, AI vision, pipeline.
Read PDFs and images as markdown or text, with exact costs and hard spend caps. $0.75/1k pages.
Generate images, GIFs, videos, and PDFs from HTML, URLs, or templates — from your AI agent.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI applications to read and process PDF files with intelligent file search, text extraction, image processing, and optional OCR support for scanned documents.MIT
- AlicenseNot gradedqualityDmaintenanceProvides random access to PDF contents with selective page extraction, text search, outline navigation, image extraction, and page rendering capabilities. Reduces token usage by allowing targeted content extraction instead of processing entire documents.4MIT
- FlicenseNot gradedqualityDmaintenanceProvides tools for extracting text, performing OCR on scanned documents, and retrieving images from PDF files. It enables Model Context Protocol clients to process PDF content page-by-page through specialized extraction tools.1-