# [Spec Title]

**Layer:** L2 Spec
**Parent Principle:** [Principle name + L1-XXX hash]
**Stake to register:** max(2 PWM, ceil($5 / TWAP_30d))

---

## Understanding This Spec — Plain Language

### What This Spec Defines
[Specific parameters: dimensions, ranges, noise models, wavelength range, etc.]

### Who Uses This Spec
[Target users and use cases.]

### When to Use a Different Spec
[Pointers to related specs for different use cases.]

---

## Technical Parameters

### Forward Model
[Explicit forward model H specific to this spec.]

### Quality Metric
[How Q_int / Q_p is computed for this spec. Include the formula.]

### Evaluation Dataset
[Specific dataset + how it's used. URL, hash, size.]

---

## Benchmarks Under This Spec

[List of L3 benchmarks pointing to this spec. Empty for new specs.]

---

## Standards Met

- [ ] Parameters fully specified
- [ ] Quality metric formalized
- [ ] Evaluation dataset specified
- [ ] Plain-language explanation present

---

## AI Authoring Notes

Notes for ChatGPT / Claude / other AI tools:

> "Fill in this spec template under principle **{parent principle name}**. My target use case is **{describe}**. Specify all parameters concretely (no placeholders). Define the quality metric as an exact formula. Identify a canonical evaluation dataset. Follow the section structure above exactly. Return the result as a complete MD file."