skills.rest
opensquilla
opensquillaOFFICIAL
Explore all skills by this author →
WebsiteGitHub

meta-pdf-intelligence

Orchestrates multi-step PDF analysis with evidence indexing and memory capture.

6.8k|540|Updated May 6, 2026
Search Tags:#analysis#comparison#evidence#PDF#Summarization#Memory# orchestration
One-click install
npx skills add https://github.com/opensquilla/opensquilla --skill meta-pdf-intelligence
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: meta-pdf-intelligence
Source: https://github.com/opensquilla/opensquilla/tree/main/src/opensquilla/skills/exp/meta-pdf-intelligence
Command: npx skills add https://github.com/opensquilla/opensquilla --skill meta-pdf-intelligence

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Analyzing one or more PDFs often requires coordinating extraction, summarization, cross-document comparison, and traceable citation, which is error-prone when done ad hoc. This meta-skill orchestrates that entire pipeline so answers stay grounded in page-level evidence.

Core Features & Use Cases

  • Request Intake & Clarification: Parses the request into a strict contract (single summary, multi-PDF comparison, or question answering) and asks the user for missing sources or questions via a structured form.
  • Extraction with Fallback: Delegates text, table, and page extraction to the pdf-toolkit skill, falling back to excerpt-only analysis when the PDF file is unavailable.
  • Traceable Synthesis & Memory: Produces an evidence matrix with IDs, page citations, and confidence levels, then stores a structured index in memory for later recall.
  • Use Case: A user pastes excerpts from two research PDFs and asks which claims conflict; the skill compares only the provided excerpts, labels the answer EXCERPT-ONLY, and saves a reusable evidence index.

Quick Start

Ask the agent to analyze or compare your PDF files or pasted PDF excerpts and answer a specific question with page-backed evidence.

Frequently Asked Questions about meta-pdf-intelligence

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

FAQPage Schema
How do I compare multiple PDF documents with an AI agent?▼

Provide the PDF paths or URLs and ask for a comparison. The intake step classifies the request as multi_compare, extracts each document, and synthesizes agreements, conflicts, and unique claims with an evidence matrix citing file names and page numbers.

How to analyze a PDF when the file is not uploaded?▼

Paste the relevant excerpts or quoted page text directly into your request. The skill detects inline excerpts, skips file extraction, and produces an EXCERPT-ONLY answer with a source excerpts table, avoiding any claims about unseen pages.

Can PDF analysis answers include page-level citations?▼

Yes. The synthesis step assigns evidence IDs (E1, E2, ...) and builds an evidence matrix with document, page, evidence, and confidence columns. Every key fact must cite at least one evidence ID.

What happens if PDF text extraction fails?▼

The workflow falls back to an inline excerpt extraction step that builds a minimal evidence packet only from filenames, URLs, and user-provided quotes. Missing page evidence is explicitly labeled as unavailable rather than fabricated.

Does the PDF analysis get saved for later sessions?▼

Yes. A traceable index step compiles documents, key facts, page references, and open questions, then appends it to memory/pdf-intel.md via the memory_save tool for later recall.

Related Skills

Explore More Education & Research Skills→
spyqsspyqs

add-pdf-reader

Extract readable text from PDF attachments in WhatsApp chats, URLs, and local files.

Community
Intermediate
ShinyGuaShinyGua

ar-evidence

Extract evidence units from reading notes and group them by theme and sub-question.

Community
Advanced
sogrimsogrim
20

catalog-search

Search Technion faculty catalogs and extract degree tracks into JSON.

Official
Advanced
jhl-labsjhl-labs
50

RAG & Vector Search

Retrieve relevant documents and augment LLM prompts with vector search.

Official
Advanced