Stage 2: Baseline Digestion

Baseline digestion bridges idea selection and experiment design.

Goal

Turn the selected base paper and its codebase into structured reference artifacts that downstream agents can consume reliably.

Primary Agent

• Scout (Gemini)

Required Outputs

All outputs live in .omr/research/baseline/:

• summary.md
• experiments.md
• method-details.md
• code-notes.md
• reproduction-checklist.md

Why This Stage Exists

Without this stage, implementation tends to drift:

• experiment settings stop matching the baseline
• engineering assumptions get invented from memory
• reproduction risks remain invisible until too late

Baseline digestion is what turns a paper into an executable reference point.