Changelog

2026-06-23

Added goal to question results for indicator popovers.
Frame-component formulas now affect API scoring when configured.
Invalid frame-component formulas fall back to average scoring and are reported through the existing top-level warnings array.
Request payloads and public metric/component privacy remain unchanged.

2026-06-14

Added lens-specific indicator_name and traffic_light fields to question results for UI dials.
Question indicators reuse weighted question scores and use organization-specific lens ranges without duplicating range definitions.
Changed the default score bands to red [-1.0, -0.2), yellow [-0.2, 0.4), and green [0.4, 1.0].
Kept metric names, components, contributor scores, weights and formulas private.

2026-06-11

Added deterministic question-goal fallback prescriptions for scored neutral/adverse questions.
Added authored low, medium, and high question prescriptions while retaining question goals as a backward-compatible neutral/adverse fallback.
Added weighted, deduplicated lens-level prescriptions governed by traffic-light urgency.
Kept target-driven AI prescription generation outside the live scoring request.

2026-06-10

Added weighted aggregate scores and classifications to lens results.
Added full question text, scoring_status, and question-based scoring_coverage.
Added organization-specific red/yellow/green ranges and mapped traffic_light signals.
Added GET/PUT traffic-light range endpoints; PUT requires org:write.
Kept metrics and metric components private.

2026-02-03

Added /v1/score with optional include_prediction SNMP forecast output.
Added /v1/lenses for selected and default lens info.
Added strict channel_type validation with helpful error payload.
Removed any metric/component exposure from API responses.
Added warnings for unsupported language codes.