AI Agent Use Cases
Views —

AI Agent Use Cases

Added today

44All 189

Prompt to Prepare a Shopping Cart from a Recipe Photo

The author reports that Grok Bot receives a recipe photo, organizes the ingredients and quantities, and creates a shopping list by excluding ingredients already available at home.

Grok Bot
Life & Hobbies

1–20 of 189

Work & ResearchWalkthrough

Agent Team That Divides Research and Chart Creation to Produce PDF Reports

The public example of the Claude Agent SDK divides the research topic, then passes the results to an analysis agent and a report-writing agent.

Claude

How it works

  1. The author enters a research topic into Anthropic's research-agent example. Before running, the README instructs the user to prepare Python dependencies and an API key.
  2. When the main agent divides the topic into subtopics, the research agent searches the web and saves notes. The analysis agent reads numbers from this file and creates charts.
  3. The report agent reads the research notes and charts to create a PDF. Tool calls and returned results are configured to be logged per session.

Results

This is an implementation example that leaves research notes, data summaries, charts, and PDFs in separate folders.

Notes

The author checked the public code and the tool setup for each role. They did not directly generate a report on a specific topic or verify the accuracy of the content.

Source

GitHub Original post ↗

Example prompt · Claude

Goal: Divide research and chart creation to make a PDF report.

Input/Environment: [Research question], [Target period], [Required sources], [Python runtime environment], [Output folder].

Reference original: https://github.com/anthropics/claude-agent-sdk-demos/tree/main/research-agent

Based on the public research-agent, divide the research request by topic and configure each agent to leave the source URL and verification time in the notes. The analysis agent must use only the numbers in the notes and specify the units and reference time. Connect the report agent so it reads the notes and chart files to create a PDF with sources. Ensure that the tool call logs allow tracking which agent used which materials.

Do not speculate about account connection methods or outcomes not present in the original; first provide the necessary setup and execution order. Start with a working copy to preserve existing materials, and mark any inaccessible items as needing verification.

Verification: Try including data from different times and claims lacking sufficient sources. Check that the chart values match the notes, that the PDF sources actually support the statements, and that incomplete research is not marked as finished.

Summarize the inputs used, execution logs, and confirmed results along with the outputs. Clearly distinguish between completed, failed, and unexecuted parts.