Module 3 Assignment: Extend the Campus Agent

Objective: Extend the DigiPen Campus Agent by adding a new specialized agent that handles a specific domain.

Background:

The Campus Agent currently has four specialized agents: Building Agent, Course Agent, Handbook Agent, and Cafe Agent. Your task is to add a fifth agent that serves a new purpose on campus.

Suggested Agent Ideas (pick one, or create your own):

• Events Agent - Helps students find information about campus events, club meetings, and activities
• IT Agent - Assists with common IT issues, lab software, and printing
• Library Agent - Helps students find resources, reserve study rooms, or check hours
• Career Services Agent - Provides information about internships, resume reviews, and career fairs
• Transportation Agent - Helps with parking, shuttle schedules, and commute options

Requirements:

1. Create a new agent with appropriate instructions (system prompt)
2. Implement knowledge retrieval using one (or more!) of these approaches:
   • Add documents to the vector store and use FileSearchTool, OR
   • Use the WebSearchTool to give your agent access to the web, OR
   • Create a custom tool using @function_tool that returns relevant data (should be more complex than the cafe example) OR
   • Create a new (or select an existing) MCP server
3. Integrate with the main Campus Agent via handoff
4. Test your agent with at least 3 different queries and document the results
5. Use the OpenAI Traces Dashboard to debug at least one interaction and include a screenshot or description of what you observed

Deliverable: A Colab/Jupyter notebook with:

• Code cells with your implementation (building on the campus-agent.ipynb starter)
• Uses OPENAI_API_KEY for the API token. (Please do not include your API key in your notebook!)
• Markdown cells explaining:
  • What agent you chose and why
  • How you implemented knowledge retrieval (vector store vs. custom tool)
  • Your test queries and the agent’s responses
  • What you learned from the Traces Dashboard (what worked, what didn’t, any debugging insights)

Bonus:

Deploy your extended Campus Agent to Hugging Face Spaces and include the URL in your submission. Remember to add your OPENAI_API_KEY as a secret in your Space settings (not in the code!).

Hints:

• Start by copying the campus-agent.ipynb notebook and modifying it
• Keep your agent’s scope focused. A narrower domain with good data works better than a broad domain with sparse data
• If using the vector store, you can upload files via the OpenAI Platform Storage
• If using a custom tool, remember that the function’s docstring helps the LLM understand when to call it