AgentSkillsCN

gemini-pdf-analyzer

借助 OpenRouter，通过 Gemini 2.5 Flash 分析 PDF 和图像。当用户需要以智能方式提取信息、提出相关问题，或对大型 PDF 或图像进行深度分析时，这款工具便能大显身手。

SKILL.md
--- frontmatter
name: gemini-pdf-analyzer
description: Analyzes PDFs and images using Gemini 2.5 Flash via OpenRouter. Use when asked to extract info, ask questions about, or analyze large PDFs or images agentically.

Gemini PDF Analyzer

Analyze PDFs and images using Google's Gemini 3 Flash model via OpenRouter API.

Usage

Run the analyzer script with a prompt and one or more PDF/image files:

bash
bun run .agents/skills/gemini-pdf-analyzer/scripts/analyze.ts "Your question here" path/to/file1.pdf path/to/file2.png

Environment

Requires OPENROUTER_API_KEY environment variable.

Capabilities

  • •Extract text and tables from PDFs
  • •Answer questions about document content
  • •Analyze images and diagrams
  • •Compare multiple documents
  • •Process large PDFs by sending them directly to Gemini's vision capabilities

Examples

Ask about a PDF:

bash
bun run .agents/skills/gemini-pdf-analyzer/scripts/analyze.ts "What are the main findings in this report?" report.pdf

Analyze multiple files:

bash
bun run .agents/skills/gemini-pdf-analyzer/scripts/analyze.ts "Compare these two documents" doc1.pdf doc2.pdf

Extract structured data:

bash
bun run .agents/skills/gemini-pdf-analyzer/scripts/analyze.ts "Extract all tables as JSON" data.pdf

Agentic Workflow

For complex multi-page PDFs, use iteratively:

  1. •First ask for a summary/overview
  2. •Then drill down into specific sections
  3. •Extract structured data as needed