Spec Reuse: Auto-load archived specs or require explicit user context? #802
Description
Description:
When iterating on an existing capability (e.g., modifying "Feature A"), how should OpenSpec handle the base specification?
Auto-Discovery: Should the framework automatically detect and load the existing/archived spec file based on the capability name to provide context for the AI?
Explicit Instruction: Must the user manually specify which previous spec file to read in every new proposal?
Goal: Clarify the recommended workflow to ensure the AI maintains context without breaking the "Single Source of Truth" principle. If auto-discovery isn't supported yet, is it on the roadmap?