[erickkkyt]

Brand workspaces can inspect AI answer evidence but lack the site's AI referral visits and crawler requests. Add an AI Traffic & Crawls page that reuses existing GA4 analytics and imports site-specific access logs, with those measures kept separate from mentions and citations.

• Connect Google through web OAuth with analytics.readonly, session-bound state/PKCE, brand-bound encrypted grants, property selection and manual refresh of the last 28 completed days.
• Show platform sessions, engagement, configured key events, landing pages and source evidence. Supplement the native AI Assistant channel with explicit platform domains and retain GA data-limit warnings.
• Import Combined/JSON/JSONL logs with hostname/size validation, declared crawler purposes, HTTP responses and bounded evidence; discard IPs, full UAs, referrers and query strings before persistence.
• Reuse organization access, existing PostgreSQL and shared UI components; include English/Chinese states and migration 0028 for one brand-scoped source table.

Validation: repository lint and all workspace type checks passed; 1,560 workspace unit tests, 49 disposable-PostgreSQL integration tests and 7 Storybook interaction tests passed. Production web build passed. Desktop and 390px English/Chinese component states were checked in Chromium without page overflow.

Operational prerequisites: configure the deployment's Google Web OAuth client, enable Analytics Admin/Data APIs and register APP_URL/api/traffic/ga/callback. Real Google consent/Data API acceptance remains pending; browser access to GA is not treated as product authorization. Existing user databases were not migrated, and no external account settings or deployments were changed.