refine

Overview

Runs an interactive refinement loop on the existing ontology. The command mirrors the tail end of ouro create by using a facilitator agent (with optional Codex auto-responses) to ask clarifying questions and update .ouro/ontology/main.md.

Usage

ouro refine [--human] [--agent-model <model>] [--codex-model <model>] [--max-questions <n>]

Key Options

• --human – Disable Codex auto-responses so you can answer the facilitator manually.
• --agent-model <model> – Override the discussion agent model.
• --codex-model <model> – Override Codex when auto-answering.
• --max-questions <n> – Limit facilitator turns (default 12).

Inputs & Outputs

• Inputs: current ontology, curated docs map (if present), and answers provided during the session.
• Outputs: updated .ouro/ontology/main.md reflecting refinements gathered during the dialog.

When to Use

• After implementing a feature that alters architecture, communication patterns, or technology use.
• During onboarding sessions to capture institutional knowledge discovered by new contributors.

Follow-up Checklist

• Re-run ouro map to reconcile detected technologies with the refined ontology.
• Commit the ontology update once the refinements are reviewed.