Objective
Refactor the current notebook experience so it reads like a guided lesson instead of a scratchpad.
Context
The existing notebook already demonstrates FastMCP, simple tools, CSV handling, commission calculations, and sample insurance data. The lesson flow should be tightened so beginners can run one cell at a time and understand what each cell proves.
Scope
- Update
notebooks/Model-Context-Protocol-101.ipynb, or split into clearer notebooks if that is simpler.
- Preserve the current sample-data learning path.
- Prefer package imports over redefining core business logic repeatedly.
- Add clear markdown before code cells.
- Strip stale execution counts/outputs if appropriate.
Suggested tutorial rhythm
Use a repeated pattern:
Problem -> MCP concept -> Code cell -> Expected output -> Business meaning
Acceptance criteria
Non-goals
- Do not build a web app.
- Do not introduce real client data.
- Do not add advanced claims triage.
- Do not require API keys.
Validation
python -m pip install -r requirements.txt
ruff check .
mypy --strict .
pytest -q
Manual validation:
- Open the notebook.
- Restart kernel.
- Run all cells in order.
- Confirm outputs match the written explanations.
Objective
Refactor the current notebook experience so it reads like a guided lesson instead of a scratchpad.
Context
The existing notebook already demonstrates FastMCP, simple tools, CSV handling, commission calculations, and sample insurance data. The lesson flow should be tightened so beginners can run one cell at a time and understand what each cell proves.
Scope
notebooks/Model-Context-Protocol-101.ipynb, or split into clearer notebooks if that is simpler.Suggested tutorial rhythm
Use a repeated pattern:
Acceptance criteria
pip install -r requirements.txt.Non-goals
Validation
Manual validation: