---

name: conditional_pair_protocol description: Build and validate SR3-style low-resolution conditioning paired with high-resolution targets for reduced or full recovery experiments.

Conditional Pair Protocol

Use this skill when a recovery experiment needs SR3-style paired data. Do not use it to evaluate image quality or to replace the diffusion objective.

Inputs

• A clean high-resolution scalar or image-like value.
• A deterministic degradation or scale factor.
• Optional provenance describing whether the item is real, resource-derived, or synthetic proxy data.

Outputs

• A JSON-compatible pair containing condition, target, scale_factor, is_proxy, and provenance fields.
• Validation errors for missing values or nonpositive scale factors.

Workflow

1. Keep the low-resolution condition and high-resolution target as separate fields.
2. Record whether the pair comes from a real dataset or from a declared reduced proxy.
3. Preserve source metadata so recovery can distinguish benchmark-derived items from synthetic examples.
4. Pass only the pair object to diffusion objective or sampler skills; do not inject evaluation decisions here.

Validation

Run python tests/test_pair_protocol.py or validate the skill tree with --run-tests.

Limitations

A synthetic scalar pair can test conditioning contracts, but it is not evidence of full SR3 visual fidelity.