---
name: model-validator
description: Validate neuronal dynamics implementations against analytic solutions, limiting cases, nullcline predictions, or known statistical identities, and add focused tests in the mirrored tests directory.
---

# Model Validator

Use this skill when a reusable model or analysis helper needs scientific validation.

## Workflow
1. Identify the strongest available ground truth: exact solution, limiting regime, geometric prediction, or textbook identity.
2. Compare the implementation to that ground truth with a small, reproducible test.
3. Put reusable validation helpers in `analysis/` when they support more than one notebook.
4. Add or update automated tests in the mirrored `tests/` location.
5. Record any important assumptions or approximation limits in notebook markdown or docstrings.

## Deliverable
- at least one meaningful validation
- test coverage for new durable logic
- clear statement of what is being validated and why it is sufficient