AI Agent Portfolio

Small Python examples that demonstrate AI-agent workflow design, task planning, tool-use concepts, evidence collection, and final summaries.

This is a compact technical project: the workflow runner is intentionally simple so the control flow is easy to inspect.

What is inside

• A minimal workflow runner
• Reusable task and tool abstractions
• Example workflows for research, planning, and evaluation
• Simple tests that document expected behavior
• JSON input examples for repeatable local runs

Project structure

agent_portfolio/
  cli.py        # CLI entry point
  workflows.py  # workflow runner and result model
  tools.py      # simple example tools
examples/
  research_brief.json
  evaluation_review.json
tests/
  test_workflows.py

Run locally

python3 -m agent_portfolio.cli examples/research_brief.json
python3 -m agent_portfolio.cli examples/evaluation_review.json

Run tests

python3 -m unittest discover -s tests -p 'test*.py'

Example workflow idea

Goal: prepare a short research brief
1. collect notes
2. identify risks or missing evidence
3. summarize next actions

Why this project exists

Agent workflows are easier to trust when the steps, tools, and evidence are visible. This project shows a small pattern for making that flow explicit.

License

MIT