PageIndex MCP
Install PageIndex MCP so your agent runs reasoning-based RAG over long PDFs from local paths or online URLs inside the build loop.
Overview
PageIndex MCP is a MCP server for the Build phase that provides reasoning-based RAG for chatting with long PDFs from local or online files.
What is this MCP server?
- Reasoning-based RAG tuned for long PDFs rather than naive chunk-only search
- Supports local files and online PDF URLs in one MCP server
- Ships as npm @pageindex/mcp 1.8.1 stdio and mcpb bundle with published fileSha256
- GitHub VectifyAI/pageindex-mcp; MCP schema version 2025-09-29
- stdio transport for local agent runs—no mandatory hosted legal-style remote
- Package and server version 1.8.1
- Two distribution channels: npm @pageindex/mcp stdio and mcpb 1.8.1 release
- Supports local and online PDF inputs per catalog description
Community signal: 356 GitHub stars.
What problem does it solve?
Dropping huge PDFs into a generic context window loses structure and recall when your agent needs reliable answers over manuals and reports.
Who is it for?
Indie builders adding PDF Q&A to agents, parsing product PDFs, or prototyping support tools over existing document corpuses.
Skip if: Teams that only need short text files, want cloud-only SaaS with zero local stdio setup, or need non-PDF office formats without conversion.
What do I get? / Deliverables
After installing the stdio MCP package, your agent can query long PDFs with reasoning-based RAG while you implement doc-aware features.
- Agent-accessible long-PDF Q&A via reasoning-based RAG
- Grounded answers over manuals and reports during feature implementation
Recommended MCP Servers
Journey fit
How it compares
Reasoning-based PDF RAG MCP server, not a general codebase search skill or legal-database connector.
Common Questions / FAQ
Who is PageIndex MCP for?
Solo developers using Claude Code, Cursor, or other MCP clients who need dependable long-PDF question answering during product build.
When should I use PageIndex MCP?
Use it in build when implementing agent features that must read long PDF specs, policies, or research papers from disk or URLs.
How do I add PageIndex MCP to my agent?
Install npm package @pageindex/mcp 1.8.1 with stdio transport or the mcpb release from VectifyAI/pageindex-mcp GitHub and register it in your MCP config.