gemini-file-analysis

Gemini File Analysis

Use this skill to send files to Gemini for multi-modal analysis. Gemini can read PDFs, interpret images, transcribe audio, understand video frames, and reason over large source files that exceed Claude's context window.

When to use this skill

• PDF documents: Research papers, contracts, architecture decision records, runbooks, changelogs.
• Images: Screenshots with error messages, diagrams, UI mockups, charts, or photos requiring visual interpretation.
• Audio files: Meeting recordings, user interviews, voiceovers that need transcription or content extraction.
• Video files: Screen recordings, demo clips, or short videos requiring frame-level understanding.
• Large source files: A codebase archive, minified bundle, or log file that is too large to fit in Claude's context.
• Binary formats: Any file format Claude cannot read as plain text.

MCP tools

Tool	Purpose
gemini_analyze_file	Upload a file and run a prompt against its contents

Usage pattern

Analyze a PDF

{
  "tool": "gemini_analyze_file",
  "arguments": {
    "file_path": "/path/to/architecture-decision-record.pdf",
    "prompt": "Summarize the key architectural decisions and list any open questions."
  }
}

Interpret a screenshot

{
  "tool": "gemini_analyze_file",
  "arguments": {
    "file_path": "/path/to/error-screenshot.png",
    "prompt": "What error is shown and what are the most likely root causes?"
  }
}

Transcribe an audio recording

{
  "tool": "gemini_analyze_file",
  "arguments": {
    "file_path": "/path/to/meeting-recording.mp3",
    "prompt": "Transcribe the key action items and decisions from this meeting."
  }
}

Analyze a large source file

{
  "tool": "gemini_analyze_file",
  "arguments": {
    "file_path": "/path/to/large-codebase.tar.gz",
    "prompt": "Identify the main modules, entry points, and any obvious code quality issues."
  }
}