pdf-reader

Extract evidence-grounded answers from local PDF pages and sections.

58|1|Updated May 13, 2026
One-click install
npx skills add https://github.com/Simplified-Reasoning/Pi-Bench --skill pdf-reader-simplified-reasoning
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pdf-reader
Source: https://github.com/Simplified-Reasoning/Pi-Bench/tree/main/data/pharmacist/skills/pdf-reader
Command: npx skills add https://github.com/Simplified-Reasoning/Pi-Bench --skill pdf-reader-simplified-reasoning

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

When key information lives inside a local PDF, it prevents you from answering accurately without manually hunting through pages, figures, legends, and methods.

Core Features & Use Cases

  • PDF-grounded extraction: Pulls only the relevant pages or sections needed for the question to reduce irrelevant context.
  • Evidence-tied answering: Links claims to page-level or section-level evidence and clearly separates direct PDF evidence from interpretation.
  • High-value follow-up: Requests the most missing, high-impact material when the PDF is insufficient instead of asking generic questions.

Quick Start

Use the pdf-reader skill when you have a local paper or figure PDF and want evidence-based answers grounded in the exact page and section.

Frequently Asked Questions about pdf-reader

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract information from a PDF research paper using evidence grounding?

Evidence-grounded PDF extraction pulls only relevant pages or sections from local PDF files to answer specific questions, linking claims directly to page-level evidence while clearly separating direct findings from interpretation.

Can I get answers tied to specific figures and legends in a PDF?

Yes, PDF figure interpretation enables you to ask questions about specific figures, legends, and methods within a paper. The extraction process targets the exact page-level content needed to justify answers based on visual elements.

What is the best way to justify answers using page-level evidence from local PDF files?

Page-level evidence justification works by selecting only the relevant pages or sections needed for a question, then tying every statement to that specific PDF evidence and distinguishing what the document states from your own interpretation.

Does PDF reading comprehension work for multi-turn workflows involving research papers?

Yes, multi-turn workflows are supported for paper reading and section-level justification. The process maintains context across turns, selecting relevant pages each time and grounding successive answers in the appropriate PDF evidence.

What happens when the local PDF lacks sufficient information to answer my question?

When the PDF is insufficient, the system requests the highest-value missing material instead of asking generic questions. This targeted follow-up identifies the most impactful absent content needed to complete the evidence-grounded extraction.