This project explores agent sdk capabilities mainly on agent triaging, handoff and tools
we have 5 different agents here
user can ask a question to generate a sql with database context or summarize, classify and sentiment analysis of document using resource context
tools will use database or file context to fetch database metadata and constraints or filecontent(here we mimic by returning some dummy data)
There is triage agent which will receive the input aand invoke the relative agent
In case of sql generation, database agent will ivoke the tool and use metadata as context along with question and return the generated query
In case of file processing content fetching agent will invoke the tool and fetch the file content
Then based on chunks returned, content processing agent will process the chunk one by one
after fetching the individual results, aggregation agent will do the aggregation of object and return the final output
To run this locally uv should be installed create .env file and copy the contents in .env_template create OPENAI_API_KEY and set it in env create venv by running uv venv install dependencies by running uv sync now you can run uv run main.py /uv run planner.py to test