AI Parse

A skill for parsing PDF files using Large Language Models.

Capabilities

• - Extract information from PDF files
• Return results in JSON or Markdown format
• Resume processing from existing task ID
• Save task ID information to JSON file for reference

Parameters

Usage Examples

Normal Upload Mode

CODEBLOCK0

Resume from Existing Task or Check Status

CODEBLOCK1

Task ID File Format

When using normal upload mode, a task ID file will be created at task_id_path with the following JSON structure:

CODEBLOCK2

This file can be used to:

• - Track the submitted task
• Retrieve the task ID later for status checking
• Resume processing if interrupted

Implementation

Implemented by handler.py which:

• - Uploads PDF files to the processing service
• Polls for processing completion
• Downloads and saves results in the requested format
• Supports resuming from existing task IDs
• Saves task ID information to JSON file

Environment Requirements

• - Python 3.6+
• requests library

Return Value

The parsed result will be saved to the specified result_path in the requested format:

• - JSON format: Structured JSON with task details and extracted content
• Markdown format: Formatted Markdown with page-by-page content

Notes

• - For large PDF files, processing may take multiple minutes
• Free users can process 30 PDF pages - visit https://api.pinocch.com/index for extra trial credits
• The --task-id parameter can be used to resume processing if interrupted
• Check the console output for processing progress and status updates
• The task ID file is created immediately after successful upload
• IMPORTANT FOR AGENTS: Before declaring a task as failed, always use the task ID to check the current status of the task. Use the --task-id parameter to resume or verify the task status. The task may still be processing or have completed successfully.