RAG Frameworks · SylphxAI

pdf-reader-mcp

PDF Reader MCP is a TypeScript-based MCP server that extracts structured content from PDFs with source evidence, visual crops, and trust reports. It integrates with Claude, Cursor, VS Code, and other MCP clients to help AI agents read PDFs reliably and cite their sources.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/SylphxAI/pdf-reader-mcp

810

GitHub stars

Forks

TypeScript

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	SylphxAI/pdf-reader-mcp
Owner	SylphxAI
Primary language	TypeScript
License	MIT — OSI-approved
Stars	810
Forks	70
Open issues	6
Latest release	v3.0.10 (2026-07-01)
Last updated	2026-07-07
Source	https://github.com/SylphxAI/pdf-reader-mcp

What pdf-reader-mcp is

An MCP server written in TypeScript that parses PDFs into an Agent Document Twin—a linked, evidence-backed representation including Markdown, JSON, HTML, page text, chunks, tables, visual evidence, OCR adapters, and accessibility/trust metadata. Supports digital and scanned PDFs with configurable OCR and visual providers via environment variables.

Quickstart

Get the pdf-reader-mcp source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/SylphxAI/pdf-reader-mcp.git2cd pdf-reader-mcp3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

AI Agent Document Processing

Equip AI agents (Claude, Cursor) with reliable PDF reading that includes page numbers, bounding boxes, visual crops, and evidence IDs so they can cite sources and avoid hallucination.

Document Intelligence for Enterprise

Extract tables, charts, formulas, and images from mixed digital/scanned PDFs with OCR fallback, trust reports (hidden text detection, prompt injection warnings), and accessibility grading.

RAG and Citation Systems

Build retrieval-augmented generation pipelines where agents can search PDFs, verify answers against visual regions, and produce source-backed citations with bounding box proof.

Implementation considerations

Node.js ≥22.13 required; install via npm (@sylphx/pdf-reader-mcp) or Docker image (ghcr.io/sylphxai/pdf-reader-mcp).
Default package operates without OCR or vision models; enable optional providers via MCP_PDF_OCR_* and MCP_PDF_REGION_ANALYSIS_* environment variables pointing to local commands, HTTP servers, Ollama, or LM Studio.
MCP client configuration varies (Claude Desktop, Cursor, VS Code, custom clients); reference the installation guide for exact setup per platform.
The 'one smart tool first' design uses auto-profiling to route extraction automatically; agents can override with explicit include_* flags for fine-grained control.
Release artifacts (json files) document benchmark passes and provider certification; inspect pdf_sota_release_gate.json and pdf_quality_benchmark.json to verify quality gates.

When to avoid it — and what to weigh

Simple Text Extraction Only — If you only need raw text and do not care about layout, evidence, trust reports, or visual verification, simpler tools like pdfplumber or pypdf may be faster to integrate.
Non-MCP Tool Integration — If your architecture does not use the Model Context Protocol, you will need to either adopt MCP or wrap this server in a custom HTTP/RPC bridge, adding deployment overhead.
Offline OCR Without Configuration — OCR for scanned PDFs requires explicit provider setup (Tesseract, Ollama, OpenAI, etc.). The default package does not include OCR models; you must supply and manage them separately.
Maximum Performance on Very Large PDFs — Node.js performance on multi-gigabyte PDFs or massive batches may lag compared to Python/C++ alternatives. Profile your document sizes and throughput requirements.

License & commercial use

MIT License. Permissive OSI-approved license permitting commercial use, modification, and redistribution with attribution and no warranty.

MIT License explicitly allows commercial use. No proprietary restrictions on bundling, resale, or enterprise deployment. However, review your cloud provider's terms if hosting this server as a managed service, and ensure compliance with any upstream OCR/vision provider licenses (e.g., OpenAI, Tesseract, Ollama licensing) if configured.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Strong
Assessment confidence	High

Security considerations

Project includes trust reports (hidden text detection, prompt injection warnings, redaction, unsafe links) and accessibility reports. MCP communication occurs over stdin/stdout, mitigating network exposure in local setups. Provider paths are deployment-controlled environment variables; no request-time provider selection, reducing injection risk. No claims of cryptographic validation or pen-test certification made in data; 'Security Validated' badge links to mseep.ai assessment (external source, not verified here). Review provider configurations for credential exposure when using cloud OCR/vision endpoints.

Alternatives to consider

pdfplumber (Python)

Mature Python library for table extraction and text coordinates; simpler if you only need text + bounding boxes and do not require MCP integration or visual evidence.

pypdf (Python)

Lightweight pure-Python PDF reader; good for text extraction and form handling, but lacks visual evidence, OCR, and agent-oriented features.

LlamaIndex/LangChain document loaders

Framework-integrated PDF loaders with vector embedding and RAG pipelines, but generic; do not provide Agent Document Twin semantics, trust reports, or MCP protocol.

Software development agency

Build on pdf-reader-mcp with DEV.co software developers

Use PDF Reader MCP to enable your AI agents to read PDFs accurately, cite sources, and verify answers with visual evidence. Install via npm or Docker. Free to use under MIT License.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

mcp markdownify-mcp mcp

Related on DEV.co

Explore the category and the services that help you build with it.

RAG Frameworks AI Application Development API Development Enterprise Software Talk to DEV.co

pdf-reader-mcp FAQ

Do I need to configure OCR to use this?

No. The default package handles digital PDFs with selectable text out of the box. OCR is optional; enable it by setting MCP_PDF_OCR_* environment variables to point to your OCR provider (Tesseract, Ollama, OpenAI, etc.).

Which MCP clients are supported?

Claude (Code and Desktop), Cursor, VS Code (with MCP extension), Windsurf, Cline, Warp, and any client that implements the Model Context Protocol. See the installation guide for platform-specific setup.

Can I use this in production with multiple concurrent PDF jobs?

Unknown—no throughput or concurrency benchmarks are provided in the data. Profile your document sizes and workload. Node.js performance on large batches may vary; consider load testing or using containers with resource limits.

Is the code secure? Are there known vulnerabilities?

Project includes trust reports and mentions a 'Security Validated' badge from mseep.ai, but no CVE list, pen-test results, or detailed security audit is provided in the data. Review dependencies, provider configurations (especially for cloud credentials), and your threat model before production deployment.

Custom software development services

Need help beyond evaluating pdf-reader-mcp? DEV.co is a software development agency offering software development services and web development for teams of every size. Our software developers and web developers build custom software, web applications, APIs, and rag frameworks integrations — and maintain them long-term.

Equip Your Agents With Reliable PDF Intelligence

Use PDF Reader MCP to enable your AI agents to read PDFs accurately, cite sources, and verify answers with visual evidence. Install via npm or Docker. Free to use under MIT License.

Talk to DEV.co Schedule a Consultation