## Objective Standardize evaluation outputs so they are reusable, comparable, and CI-friendly. ## Scope Create a stable JSON results schema and Markdown summary format. ## Tasks * Define results schema including: * schema_version * run_id * timestamp * bundle_name * git_sha if available * per-item metrics * aggregated metrics * pass_fail indicators * Implement JSON result writer * Implement Markdown summary generator * Add threshold comparison logic * Ensure deterministic output formatting ## Acceptance Criteria * Running evaluation produces: * results.json * summary.md * Schema includes version field * Exit code is non-zero if thresholds fail * Aggregated metrics computed correctly ## Out of Scope * Azure Monitor export * KQL integration * Dashboard publishing
Activity
[-]Epic D - Normalized Output Schema[/-][+]Normalized Output Schema[/+]placerda commentedon Feb 25, 2026
Combined with another item for clarity.