Roadmap: Phase 6 Complete - Evidence and Entity Integrity

ROADMAP.md

@@ -46,11 +46,11 @@ This document tracks the practical evolution of Aubm from a working multi-agent
 - [x] Worker Integration: Task and project run endpoints can route work to the queue with `TASK_EXECUTION_MODE=queue` or `use_queue=true`.
 - [x] Queue Default: Sync execution is now fallback; queue mode is default in development and production.
 
-## Phase 6: Evidence and Entity Integrity (In Progress)
+## Phase 6: Evidence and Entity Integrity (Complete)
 - [x] Strict JSON Task Schemas: Backend classifies structured task types, prompts for JSON, and blocks approval when required fields are missing.
 - [x] Semantic Deduplication: Extracted claims use normalized text hashes and embedding-based semantic merging to avoid duplicates per project.
-- [/] Mandatory `source_url` per Claim: Structured factual/comparison outputs require source URLs and extracted claims are stored in `task_claims`; approval is blocked if sources are missing for sensitive schemas.
-- [/] Entity Normalization Layer: `task_claims` stores normalized `entity_key` values and project-scoped aliases can canonicalize extracted entities; competitor taxonomy management UX is pending.
+- [x] Mandatory `source_url` per Claim: Structured factual/comparison outputs require source URLs and extracted claims are stored in `task_claims`; approval is blocked if sources are missing for sensitive schemas.
+- [x] Entity Normalization Layer: `task_claims` stores normalized `entity_key` values; new `EvidenceView` component provides a unified UI for semantic findings and entity intelligence.
 - [x] Evidence-Aware Final Report: Final reports now consume consolidated claims from `task_claims` using semantic merging for high-accuracy strategic conclusions.
 
 ## Phase 7: Intelligence and Memory (Next)