Ansvisor Files

Added

• Query Fan-out: the observed sub-queries answer engines actually run while building your answers are now captured and surfaced. Tracking stores the fan-out returned by Cloro (prompt_results.search_queries, no extra scraper spend), and a new Query Fan-out tab on /dashboard/prompts lists the distinct sub-queries with how often each was searched, the source engines, the prompts they came from, and a one-click "+" to promote any of them into a tracked prompt (#341, [#342])
• Citations: Competitor Gaps — a view of domains that cite your competitors but not you, with a per-competitor source map so you can see exactly who's earning those citations (#326, [#327])
• Citations: add a cited domain as a competitor inline with a "+" straight from the Domains table (#329)
• Citations: hover any prompt in the prompt filter dropdown to read its full text (#324)
• Prompts: sortable columns on All Prompts (visibility / mentions / volume / last run) with the active sort deep-linked in the URL (#322)
• Prompts: a banner surfaces keywords that haven't had volume analysis yet, with a one-click Analyze action (#296)
• Brands: pause a brand — a paused brand keeps all its history but is skipped by the daily tracking cron and on-demand runs, so it spends no Cloro/LLM credits until resumed (#308)
• MCP: run a Site Audit from any MCP client (run_site_audit + get_site_audit, quota-charged), plus list_site_audits and get_site_audit_quota read tools (#305, [#306])
• Assistant: the in-product AI assistant can now run a Site Audit itself, and list_citations gained a source_filter so it can isolate owned / competitor / external citations (#307, [#311])
• Traffic: only platform-attributable AI visits (a real AI referrer or a known utm_source) are persisted now — unattributable "unknown" hits are dropped (#310)
• Billing: the free trial is now 14 days, up from 7 (#319)
• Insights: the region filter shows readable region names instead of raw codes (#315, thanks @Peter7896)

Changed

• Server: structured logging Phase 2 — the remaining console.* calls across the routes (request-scoped req.log), workers, job manager/runner, server.js, middleware, and lib/config were migrated to the pino logger, so production logs are leveled, JSON-formatted, and correlatable by request id (#330, [#334], [#335], [#336], [#337])
• Server: consolidated the intent-extraction logic into a single shared module (#302)
• Web: centralized the API base-URL configuration so every server action resolves it one way (#292, thanks @BharadwajKanneveti)
• Content Optimization: reordered the KPI cards so Avg. Score comes before Sent to Workflow (#321)
• Insights: the metric breakdown rows are now sorted by their current value (#320, thanks @Peter7896)
• Self-host: ship a generated consolidated supabase/schema.sql so a fresh install can be created from one file instead of applying every migration by hand (#331)
• Brands: removed unused imports in the brand settings page (#323)

Performance

• Topics: opening a topic is much faster — the detail page's six serialized server-action calls were collapsed into one action that runs the reads in parallel server-side (#338)
• Citations: switching between the Domains and URLs tabs is now instant (keepMounted + memoization) (#325)

Fixed

• Team: the "share this link directly" invite link now points at the self-contained /invite/{token} accept page instead of an email-only /auth/confirm URL, so a shared or copied invite no longer 404s to ?error=auth_confirm_missing_params; re-inviting a removed member works end to end, and the invite dialog is honest about whether an email was actually sent (#340)
• Billing: non-admin members are blocked from the payment/onboarding screens when an organization's subscription lapses (only admins settle billing) (#317)
• Billing: trialing subscriptions are treated as active in the web plan-guard, so trial users aren't gated out of paid features (#285)
• Insights: the visibility score is no longer floored to 0, so sub-1 averages stay visible (#283)
• Site Audit: AI fix recommendations are anchored to the current date instead of drifting to a stale one (#284)
• Traffic: the tracking beacon is sent as text/plain so it stays a CORS-safelisted request and isn't silently blocked (#287)

Docs

• README: added the AI Visibility Glossary to Resources (#318) and listed Scrape.do under Optional Services for the Site Audit page fetcher (#295)
• Documented SCRAPEDO_API_KEY and AUDIT_LLM_MODEL for Site Audit in the env references (#294)

Contributors

Thanks to everyone who contributed to this release, including @Peter7896 and @BharadwajKanneveti! 🙌