Skip to content

Open-source agentic AI Search Execution Platform. Track AI visibility, find opportunities, and let agents turn signals into prioritized actions and drafted fixes across ChatGPT, Claude, Gemini, Google AI Overviews, Google AI Mode, Perplexity, Grok, and Microsoft Copilot.

45.4 KB
Raw Blame

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[0.3.0] - 2026-08-24

Added

  • Google Analytics 4 integration — connect a GA4 account from Settings → Integrations through Composio managed auth, map each brand to one of its properties, and sync AI-sourced sessions, conversions and revenue daily (#695, #703, #718). AI Traffic then reports what AI referrals are actually worth, not just how many arrived, and instances without Composio credentials degrade to a safe no-op
  • Prompts can target several locations at once — a prompt's tracking locations are now a per-prompt list of countries and US states (US-CA), edited from the prompt card with a searchable picker (#783). Plan quota meters the tracked location rather than the prompt row, since a prompt tracked in three places costs three times the scraper and model calls; every write path measures the change it makes, so re-saving at the cap succeeds and removing a location frees capacity immediately (#781). Google AI Overview and AI Mode have no sub-country mechanism, so their state locations collapse to one country-wide run instead of billing the identical query twice
  • Per-prompt visibility broken down by location — a Location control on the Prompts page re-reads the health columns for one tracked location, so a prompt that is strong at home and invisible abroad no longer hides behind a blended average (#784)
  • Pages AI engines ignore — analytics-backed detection of valuable pages that AI search sends no traffic to, surfaced in AI Traffic and turned into prompt suggestions built from the pages worth defending (#724, #725, #720)
  • Prompts: suggestions get their own tab that shows what feeds them (#727), a new prompt is analyzed the moment it is added (#754), and a 7d / 30d / 90d range selector (#697)
  • One date range control across Visibility, Citations and Prompts, so the three pages can no longer disagree about the window they describe (#722)
  • Tracking: an OpenRouter provider, with sentiment analysis made provider-agnostic (#708)
  • Reports moved into the sidebar (#750)
  • Ops: CI flags pull requests that change a migration, so the apply step is not forgotten (#736)

Changed

  • Citations, Topics and Insights aggregate in Postgres instead of downloading their windows. The overview pages stopped shipping tens of thousands of rows to the browser to add them up (#726, #744); the per-URL detail page and Competitor Gaps read the citation index instead of paging every answer, which also removed a silent 50,000-row truncation that had been hiding the newest answers from both (#776, #778); and the dashboard serves from daily rollups rather than rescanning history (#773). Citations are written as rows alongside each answer to make this possible (#737, #739)
  • Content Optimization generates three opportunities per run instead of five to fifteen, so a run produces work a team can actually act on (#735)
  • Prompts: the All Prompts table sorts by visibility score descending by default (#714)
  • Citations: source filtering became a segmented control beside the table it scopes, replacing the server-side source filters (#742, #748)
  • Relative-time formatting is resolved through one shared helper instead of per-page copies (#767, #772)

Fixed

  • Security: the citation RPCs enforce organization membership. Six security-definer functions could be called with any brand UUID by any authenticated user, leaving the tenant boundary to the web tier alone; they now check membership themselves (#779, #780)
  • Settings: organization updates are restricted to admins, with server-managed columns locked against client writes (#755)
  • A brand's first day no longer sends two Daily Pulse emails — the catch-up sweep treated the deliberately unpulsed onboarding run as a missed pulse (#774), and a pulse can no longer be sent twice for the same tracking window (#717)
  • Tracking runs survive slow and stalled Cloro queues — runs whose first result lands late are no longer discarded (#710), ghost tasks no longer hold a run open (#690) or make a watched run sit through the full thresholds (#761), the pending-task read is paged so large runs aren't cut short (#716), and Grok submissions stop while the provider returns 500 for all of them (#734)
  • Citations: answers with many citations are no longer dropped by unchunked URL lookups (#738), the stored lookup pages past PostgREST's 1000-row cap (#740), URLs resolve through the request body instead of an over-long query string (#741), "has any citations" is answered without counting every row (#707), content stays visible during refetch (#775), an ambiguous column reference is resolved (#743), and a load failure no longer toasts after leaving the page (#696)
  • Prompt volumes: a brand's keywords are analysed when its setup finishes rather than only after checkout (#765), volume analysis covers a whole brand instead of failing past 50 prompts (#747), DataForSEO is asked once per brand instead of once per prompt (#749), and progress is reported from the run and picked up again on reload (#751, #752)
  • Reports: sections aggregate in Postgres and a report survives one failing section (#762); query fan-out ranks over the whole window and shows per-engine coverage (#764)
  • Content Optimization: filters stay reachable on zero results and selections can be deleted (#733); metrics are calculated across all opportunities (#700); the empty-state Generate button respects the cloud restriction (#699)
  • Topic detail trend charts and reports use the AI Visibility Score formula (#685, #693)
  • Slugs transliterate non-ASCII letters instead of deleting them (#763)
  • CSV: fields containing lone carriage returns are quoted (#758)
  • Brand avatars fall back instead of rendering nothing — a manually-set logo URL, then the primary domain's Google favicon, then the site's own /favicon.ico, then initials. Google answers an unindexed domain with a 16×16 grey-globe placeholder rather than a failure the browser reports, so it is detected by size instead of waiting for an error that never fires (#759, #782). Brand Settings gains a Logo URL field for the manual override, and a derived favicon is no longer written to logo_url at brand creation or on a domain change — the column now means "the icon this brand chose", which is what lets the chain run at all
  • Mobile navigation preferences are aligned with the sidebar (#771)
  • Database: the missing cloro_pending_tasks migration was added (#692) and the prompt_suggestions table definition recovered (#711)
  • CI: messages/*.json is covered by the format check (#698)

Docs

  • The README is retitled and rewritten around AI Search intelligence (#769, #770)
  • Migration 00055's comment is corrected — there was no schema drift (#723)

Contributors

Huge thanks to everyone who shipped code in this release — and a special welcome to our first-time contributors: @mikam44700, @shahzainshafique, @dawNotPoi, @luozejian and @kn-dev-code 🎉

Thanks as well to our returning contributors @BharadwajKanneveti, @Maqbool61 and @Pavel-glitch-ui for coming back with more! 🙌

[0.2.0] - 2026-08-09

Added

  • Daily Pulse — a per-brand digest generated after the daily tracking run: KPI strip (visibility rate, 7-day trend, mentions, citations, sentiment), highlights (first-time citations, top prompt gains, leaderboard overtakes, first appearance on a new engine) and anomaly warnings (sharp visibility drop, competitor surge, high-volume prompt losing citations) with a platform-outage guard and a 7-day warning cooldown. Email via Resend on cloud (active/trialing orgs only), a daily_pulse.created webhook event everywhere, and Settings → Notifications for per-brand frequency and recipients (#540, #561)
  • Google Search Console integration — connect a Search Console account from Settings → Integrations via Composio managed auth, map each brand to one of its properties (unambiguous domain matches are premapped automatically), and sync 28 days of query stats daily with 90-day retention (#577, #642, #643, #644, #647). Prompt suggestions are then fed by real search demand: queries the brand actually ranks for but doesn't track yet, filtered for coverage, split into head and long-tail, badged by click behaviour (protect traffic / capture demand / low competition) and enriched with cached competition data — GSC-sourced suggestions carry their source query, impressions and clicks (#648, #650). Brands without a connection keep the existing suggestion flow byte for byte, and instances without Composio credentials degrade to a safe no-op
  • Citations: per-URL detail page — every cited URL opens onto the answers that cited it, the prompt breakdown behind it, and for brand-owned URLs a bridge to its targeting and traffic (#535, #556), with a paginated prompt breakdown table (#600, #628)
  • Prompts: fan-out coverage in the By-prompt view — each prompt now shows how many of its tracked answers triggered a live search (12 / 500 · 2%) alongside the sub-query count, so prompts the engines never search for finally appear as 0 / N instead of vanishing from the table (#543, #651)
  • Insights: visibility rate trend chart with a line per tracked brand and end-of-line logos (#571); a Formulas dialog in the header explaining every metric (#581); a Reports shortcut next to Export CSV (#567)
  • Tracking: optional US state-level geo-targeting for prompts (#579), and brand/competitor mention positions are recorded on every result to power the position component of the visibility score (#569, #572)
  • Topics: sortable leaderboard columns (#620)
  • Exports: CSV for the AI Traffic visit log (#607, #630) and for Competitor Gaps, with an explanation for the intentionally disabled Source Types export (#601, #629)
  • Prompts: a citations column on the All Prompts table (#534), engine icons in the fan-out queries table (#542), a search box on the fan-out High frequency view (#593, #615), and pagination plus search on Top Sources (#530, #533)
  • Content Optimization: server-side pagination on the opportunities list (#610, #633), and search now queries the whole filtered set server-side instead of only the loaded page — so a match on page 3 stops reading as "no results" (#664, #681)
  • Billing: prompt volume analysis is triggered automatically after checkout (#582)

Changed

  • Visibility Rate is now the AI Visibility Score — a 0-100 blend of how often AI answers mention the brand (60%), cite its site (25%) and how early it's named among tracked brands (15%), computed identically on every surface: Insights KPI header and trend chart, the leaderboard, All Prompts column and prompt detail, Topics overview and detail, Competitors cards, Reports (new reports; older snapshots keep rendering their stored values), the Daily Pulse email, MCP, the public API and the Looker connector (#573, #574, #575). Coverage (appeared in X of Y prompts) stays visible as a secondary line, and detection remains fully deterministic — no LLM judges
  • Prompts: the All Prompts Visibility column now shows the prompt-level Visibility Rate (share of runs in the last 30 days with a brand mention or citation) instead of the all-runs average score, which runs without any brand presence diluted into misleading single digits — same rate language as the Insights headline and topic pages; the tooltip keeps run counts and the average score across visible runs, and the CSV export gains visibility_rate_30d / visible_runs_30d (#562)
  • Insights: the 24h view anchors to the last completed tracking run instead of a wall-clock window, so the KPI header, the dashboard and the Pulse email always describe the same slice (#578); the trend chart plots the rolling selected window so it can never disagree with the headline (#576)
  • Reports: prompt and topic performance ranks by visibility rate rather than the diluted average (#568)
  • Metrics: the position factor is shrunk by mention evidence and its full support raised to 20 mentioned answers, so a single lucky first mention no longer reads as dominance (#584, #585)
  • Competitors and topic detail surface the prompt-level visibility rate (#493, #555)
  • Platform display labels are resolved from one shared map instead of per-page copies (#553)
  • Prompts: pagination unified with the shared TablePager (#544, #546)
  • Ops: the cloud daily-tracking cron moved to 02:00 UTC (05:00 TR) — self-host schedules are unchanged and stay configurable through DAILY_CRON_SCHEDULE; the server image defaults NODE_ENV to production, and CI now runs a full Next.js production build (#528)
  • Insights: unused chart components removed (#589, #625)

Fixed

  • Tracking runs tolerate delivery gaps and refuse to stamp partial runs — the stall window is configurable (CLORO_STALL_POLL_LIMIT) and a run that produced far fewer results than it planned no longer becomes the 24h anchor, which previously let a half-delivered night rewrite the dashboard and the Pulse email with numbers nobody could reproduce (#649). Run result counts are read back from the database when stamping, so scraper-only brands no longer stamp zero (#583)
  • Pulses lost to process death are recovered by a daily catch-up sweep: any brand whose latest completed run never produced a pulse gets one dated to that run's day, with double-sends made impossible by the existing unique key (#654). The pulse also waits for the brand's scraper queue to drain before computing, so the email can't report a partially-delivered window (#570), and PUBLIC_APP_URL is normalized to an origin so email links stop pointing at doubled paths (#639, #641)
  • Content Optimization: opportunities and briefs are generated in the brand's language instead of always English (#638, #640); dismiss actions are gated behind a confirmation dialog for bulk and an undo toast for single rows (#611, #653); stale responses can no longer overwrite the list on rapid filter changes (#663, #682)
  • Site Audit: a persistent error card with retry replaces the blank page on load failures (#666, #678); re-run is guarded against double clicks that started two audits and spent two units of quota (#667, #677); the trend chart groups by local day so timezone-shifted audits stop producing duplicate labels (#668, #675); deleting an audit refreshes the trend card and quota (#669, #674); polling timeouts render a real state (#612, #632)
  • Error states with retry replace silent failures or misleading empty states across Citations (#599, #623), Topics (#595, #618), prompt detail (#591, #627), prompt suggestions (#609), invoices (#559), agent actions (#551, #646), prompt workflow validation (#549, #645), invites (#566), savePromptSet (#536) and addPromptToSet (#580)
  • Citations: the Region filter is populated from observed data (#598, #622) and YouTube video IDs survive URL normalization (#558)
  • Topics: renaming a topic no longer zeroes its prompt count — prompts are counted by topic_id instead of category name (#594, #617); the CSV export writes the real topic_id instead of an always-empty column (#590, #626); KPI strings are pluralized with aligned decimal formatting (#619)
  • AI Traffic: chart date labels are correct across timezones (#605, #621), platforms with zero visits are filtered out of the breakdown (#616), and the empty state is period-specific (#608, #631)
  • Site Audit hub: the empty state no longer flashes while loading, and failures or a missing brand no longer render as "no audits yet" (#613, #624)
  • Accessibility: insights tooltips and legends are keyboard/screen-reader reachable (#588, #604); the audit range toggles, URL input and signal panels expose their state (#671, #672); Content row actions and the search input have accessible names (#665, #679)
  • i18n: the audit hub and report render from message keys instead of hardcoded strings, including status labels that previously leaked raw enum values (#670, #673)
  • Formatting: mention/citation counts use thousands separators (#603) and the metric breakdown sheet pluralizes day and mention counts (#602)
  • Insights: the Share of Voice pie tooltip explains both of its numbers (#565)
  • Competitor aggregates drop rows for deleted competitors (#560)
  • Agent: max output tokens raised so long answers stop truncating (#563)
  • API: the rate limiter keys on the auth token instead of the client IP, so users behind a shared NAT stop throttling each other (#537)

Docs

  • A custom robots.txt with explicit AI-crawler allowances and a sitemap reference (#634)

Contributors

Huge thanks to everyone who shipped code in this release — and a special welcome to our first-time contributors: @arambu1a, @Adarshhic, @Pavel-glitch-ui, @marceli1404, @Jesulac, @MFA-G and @abdulm5 🎉

Thanks as well to our returning contributors @d180, @BharadwajKanneveti, @abdullah91111 and @Maqbool61 for coming back with more! 🙌

[0.1.7] - 2026-07-25

Added

  • Reports: templates with a section picker — four ready-made templates (Weekly Visibility Summary, Monthly Executive Report, Competitor Benchmark, Citation & Sources Report) pre-select a section set the user can toggle per report; four new modules, previous-period deltas and a change-story executive summary (#419), plus mention & citation evidence sections (#444)
  • Public metrics API v1 under /api/v1, with docs (#422)
  • Looker Studio Community Connector (#426)
  • Insights: Visibility Rate is the headline metric — the prompt-level rate (appeared in X of Y tracked prompts) leads the KPI header and the competitor leaderboard (#490); period-aware Tracked Prompts KPI card with the plan quota (#471); Topic & Prompt Opportunities teaser cards (#459, #479); a refetch overlay while filter changes reload data (#494); head-to-head links on the leaderboard (#506); a no-competitors teaser pointing to the Competitors page after its sidebar entry was removed (#504, #507, #519)
  • Prompts: prompt workflow — a work status (todo / in progress / done), a notes thread, and target URLs with automatic "cited" tracking as new results arrive, surfaced on both the detail page and the All Prompts table (#511); Add Prompt dialog on the All Prompts tab (#483); edit/delete a prompt in place (#484, #487); Top Sources card (Domains/URLs tabs) on the prompt detail page (#495) with an Insights-style date-range filter (#496)
  • Topics: persisted AI topic suggestions — generated from prompt-data gap signals, stored with accept/dismiss, plan-gated refresh (#463, #497); the overview table, Best/Weakest KPIs and CSV export lead with the prompt-level Visibility Rate (#493, #499); topics can be created directly from the Topics page (#522)
  • Citations: CSV export for Top Domains / Top URLs (#449); Source Types moved into a full-width sources-card tab (#486, #488)
  • Traffic: pagination and filtering for AI referral visits (#454); a 7/30/90-day range picker (#481), upgraded to a full date-range picker (#509)
  • Billing: Stripe Customer Portal access from Settings → Billing (update payment method, billing address and VAT/tax ID), and checkout now collects billing address and tax ID up front (#524)
  • MCP: topic suggestion tools — list, accept and dismiss (#513)
  • Plans: Claude tracking is gated to Enterprise per-customer opt-in on cloud (#423)
  • Onboarding: the plan step clarifies that the card is not charged until the trial ends, with a Stripe trust line (#503)
  • Sidebar / Mobile nav: Agent nav item shows a "Set up" badge on cloud when the org has no Anthropic API key saved; self-host instances are unaffected (#456, #472)

Changed

  • Sidebar: Prompts moved above Topics in the Analytics group (#531)
  • Models: Claude tracking upgraded to claude-sonnet-5 (#421); the duplicated model display-name maps were centralized in one config (#501)
  • Prompts: suggestions moved to the All Prompts tab (#466) and collapse to a one-line strip by default — zero suggestion requests until expanded (#470); Query Fan-out opens on the By Prompt view by default (#453)
  • Insights: removed the Prompt Results by Topic tree (#458, #477)
  • Citations: the default date preset is now 24h, with a period-aware empty state (#485, #489)
  • i18n: the Insights (#518), Topics (#517) and Agent (#516) pages are wired to next-intl

Fixed

  • Billing: monthly quotas anchor to the subscription billing period, not the calendar month — an org renewing mid-month gets its allowance back at renewal (#500)
  • Insights: Positive Sentiment divides by brand-mentioning answers — the score is no longer diluted by answers that never mention the brand (#508, #512); zero-base deltas render as "+N new" instead of a meaningless percentage (#520, #521); brand and competitor averages share the same denominator (#478)
  • Tracking: Perplexity query fan-out survives the upstream switch to a plain string-array search_model_queries — the parser now accepts both the new string entries and the legacy object shape (#525)
  • Reports: the KPI section leads with the visibility rate, with a fallback for old snapshots (#515); the AI executive summary reads *Change fields as percentage deltas and never implies an impossible previous value (#514, #520, #521)
  • Citations: citation counts are hostname-based with an automatic recount when brand domains change (#433); chatgpt-shopping rows are excluded to match the Insights KPIs (#434); overview and gaps scans page past the silent 1000-row cap (#430)
  • Topics: the overview scan paginates and excludes chatgpt-shopping (#465)
  • Traffic: exact visit counts with paginated breakdown/trend scans (#468)
  • Prompts / Query Fan-out: complete aggregation with bounded fetches and real error messages (#428); intents classify progressively — visible page first (#448); the High frequency pager is windowed (‹ 1 … n-1 n n+1 … last ›) instead of one button per page (#446, #455); suggestion refresh no longer fails on high-volume estimates (#480); the page's first load is consolidated into one server action (#473); tab/sort URL params update shallowly without a full navigation (#474); loading spinner styles aligned (#482)
  • Onboarding: LLM suggestion calls retry and fall back to manual entry (#443); Back/Continue no longer duplicates brands (#467)
  • Shopping: product prices are rounded and formatted (#418); chatgpt-shopping stays off brands that have Shopping disabled (#432)
  • Content: generated opportunity titles/descriptions no longer leak prompt indexes (#442)
  • i18n: template name lookups are guarded with t.has() to prevent MISSING_MESSAGE crashes (#425); the proxy bypasses next-intl for RSC and server-action requests (#475)
  • MCP: prompt-level visibility rate exposed and the competitor average denominator fixed (#491, #502)
  • Web: all outstanding ESLint warnings resolved (#437, #526)

Docs

  • Platform guide for Reports — templates, section picker, PDF export (#438, #527)
  • README: Blog added to Resources; website links cleaned up

Contributors

Thanks to everyone who contributed to this release, including @Maqbool61, @d180, @chinmaywadhe26, @abdullah91111, @Dodothereal, @BharadwajKanneveti, @Utkarsh-Singhal-26 and @Sam-syntax35! 🙌

[0.1.6] - 2026-07-12

Added

  • Reports (Simple Reports MVP): generate an immutable snapshot report for a brand over a chosen period — AI-written executive summary, KPI cards with deltas, visibility trend chart, share of voice, competitor leaderboard, best/weakest prompts, observed query fan-outs, and top citation sources — plus a report library and a true vector PDF export (selectable text, smart page breaks, embedded Inter so Turkish characters render correctly). Soft-launched: reachable at /dashboard/reports, no sidebar entry yet (#390)
  • Prompts / Query Fan-out: Intent column — sub-queries are classified once via LLM and cached, so intents load on demand everywhere after (#333, #353)
  • Prompts / Query Fan-out: By Prompt grouped view — expand any tracked prompt to see the sub-queries its answers actually ran (#358)
  • Prompts / Query Fan-out: the high-frequency table is paginated at 10 rows per page with numbered controls (#349, #355)
  • Insights: the result detail now shows the query fan-out captured for that answer (#362)
  • Content: the opportunity's Source Data card shows the related prompt's observed query fan-outs — first-hand demand signal next to the estimated keywords (#392, #405)
  • Content: bulk Done button for selected opportunities (#356), and opportunities are generated in the brand's language (#369)
  • Agent: completed Site Audit results render inline in the chat as an audit card (score gauge, category breakdown, top recommendations) instead of raw JSON (#272, #386), and assistant messages gained a copy button (#371, #374)
  • Shopping: provider logos replace raw platform slugs in the product tables (#398, #401)
  • Citations: the Top Sources Domains and URLs tables are paginated at 100 rows per page with globally continuous ranks and per-tab page state (#395, #411)
  • Teams: removed the member seat limit on cloud plans (#373)

Changed

  • Sidebar: "Answer Engine Insights" is now Visibility and "Content Optimization" is now Content (#391)
  • Insights: the page title is wired to next-intl (part of #384) (#389)
  • Prompts: removed the misleading Keywords and Multiplier columns from the Prompt Volumes table (#347, #359), and large counts use the shared compact number formatter (#348, #370)

Performance

  • Insights: the initial load no longer scans the full results table — counts and limits moved into SQL (#313, #352)
  • Citations: domain and article-type classifications are memoized per load (#366, #375)

Fixed

  • Shopping: ChatGPT Shopping cards were captured but never reached the dashboard — the normalizer had no ChatGPT branch, so only Copilot cards ever showed. Added the parser (derived from real captured payloads), fixed thousand-separator price parsing ("₺2.699,99" no longer reads as 2.699), and history is recoverable via the existing backfill script (#399, #408)
  • Shopping: filter selects show readable labels instead of raw values (#396, #397)
  • Onboarding: the competitor step can no longer be skipped without adding at least one competitor (#377, #378)
  • Content: sending to a workflow with none connected shows a short "No workflow connected" error instead of a redacted production stack message — single and bulk send (#393, #400)
  • Content: filter selects show their labels in the trigger (#364, #368), and the list no longer flashes a skeleton on filter changes or bulk actions (#367)
  • Prompts: fan-out intent badges paint together with the table instead of popping in (#380, #387); tracking a fan-out query updates the row in place without a skeleton flash (#346, #407) and reflects on All Prompts (#357); the unanalyzed banner says prompts, not keywords (#381, #388); navigating away mid-load no longer fires a spurious "Failed to load prompt data" toast (#394, #402)
  • Topics: KPI cards no longer render a bare 0pts delta when the change is zero (#360, #410)
  • Citations: the competitor filter shows the competitor's name in the trigger instead of its id (#344)

Docs

  • README and docs: Ansvisor mentions, the docs topbar logo, and managed-cloud references now link to the website (#376)

Internal

  • Server: structured logging final slice — the Cloro scraper joined the pino migration (#351)

Contributors

Thanks to everyone who contributed to this release, including @Sam-syntax35, @Maqbool61, @d180, @prakashiitp, @BharadwajKanneveti and @AyushSrivastava1818! 🙌

[0.1.5] - 2026-07-06

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! 🙌

[0.1.4] - 2026-06-21

Added

  • Site Audit: AEO/GEO page scoring under Content Optimization — fetches any URL (Scrape.do proxy + JS render) and scores it across 47 weighted signals in five categories (structure, content, authority, E-E-A-T, trust) using deterministic evaluators plus a batched LLM pass, then returns prioritized, AI-written fix recommendations. Runs asynchronously with live progress, a per-audit detail page at /dashboard/audit/[id] (re-run + delete), a primary-domain score trend and category breakdown on the hub, and a monthly per-plan quota (#259, #261, #262, #263, #264, #265, #268)
  • Server: structured logging — a pino-based logger with levels (LOG_LEVEL), per-request correlation IDs (x-request-id + per-request child loggers), JSON output, and sensitive-header redaction; the per-request access log dropped to debug so it's off by default in production (#273, thanks @Pallavikumarimdb)
  • Citations: expanded the source-category domain lists so more citations classify into the right bucket (#276, thanks @BharadwajKanneveti)

Changed

  • Performance: dashboard charts (Recharts) are now lazy-loaded via next/dynamic with skeleton fallbacks, trimming the initial route JS across Insights, Shopping, Citations, Topics, Traffic, Prompts, and the Agent panel (#281, thanks @BharadwajKanneveti)
  • Insights: the date range now defaults to the last 24h instead of all-time (#274, thanks @Srija-65)
  • Insights: clearer "Queued — starting automatically" copy when an analysis is waiting behind another run (#280)
  • Web: dropped the unused framer-motion dependency (#266) and removed a deprecated unused Project type (#258) (both thanks @BharadwajKanneveti)

Fixed

  • Tracking: the "Analyze Prompts" action no longer re-analyzes prompts that already have results — closes a double-spend where the same prompts could be submitted several times during the async webhook window (#278)
  • Tracking: the analysis progress bar no longer freezes partway on webhook-mode runs — the drain loop now counts only the current run's tasks (not brand-wide orphans), gives up early if delivery stalls, and a periodic sweep clears orphaned pending-task rows (#279)
  • Shopping: Microsoft Copilot shoppingProducts wrappers are flattened into per-product cards instead of a single "Unknown Product" (#255)
  • Web: switching brand tabs no longer flashes a spurious "Failed to fetch" toast from the aborted in-flight request (#257, thanks @gitbasitmalik)

Tests

  • Server: closed the remaining server-side test gaps tracked in #125 (#256, thanks @Pallavikumarimdb)

Docs

  • Added CRON_SECRET to both .env.example files, with a note that it's cloud-only and must match on the web app and the server (#277, thanks @P-Maheswari)

Contributors

Huge thanks to everyone who contributed to this release — and a special welcome to first-time contributors @gitbasitmalik, @Srija-65, and @P-Maheswari! 🎉 Thanks also to @Pallavikumarimdb and @BharadwajKanneveti. 🙌

[0.1.3] - 2026-06-14

Security

  • Internal API routes now enforce org/brand ownership on every request — closed a set of IDOR gaps where a :brandId / :id / :jobId in the URL was trusted without checking it belonged to the caller's organization (tracking, content, and volumes routes) (#246)
  • Enabled Row Level Security on previously exposed tables: jobs and prompt_volumes (server-only, no client policy) and competitors / topics (org-membership-scoped member policies mirroring content_opportunities) (#250)
  • The on-demand tracking endpoint (POST /api/tracking/check) now goes through the same cloud cost guard as analyze-new — inactive subscriptions get 402, daily-cap / cooldown get 429 — so it can no longer bypass quota on cloud (#252)
  • Cloro callback (/cloro/callback) now verifies the webhook signature before processing (#229)
  • Aggregate / row-fetch RPCs flipped to SECURITY INVOKER so they run with the caller's RLS context instead of the definer's (#200)
  • RBAC: write controls on Manage Prompts / Manage Topics are hidden for non-admin/manager roles, and Settings → Agent Save/Remove is gated behind admin (#141, #142)

Added

  • Shopping: end-to-end Shopping suite — brand-level Shopping mode toggle, ChatGPT Shopping platform, normalized prompt_result_shopping_cards with a parser worker, sidebar entry + overview page, My Products / Competitors tabs with brand matching, a card-eligible prompts tab, and Insights isolation (#143, #144, #155, #157, #176, #178; #176 and #178 thanks @Pallavikumarimdb)
  • Agent: render_chart tool with inline Recharts visualizations in the chat panel (#138)
  • Content: monthly quota for content brief generation (#224)
  • Citations: "Own domain only" filter to isolate first-party citations (#164, thanks @Pallavikumarimdb)
  • Auth: password visibility toggle on the auth forms (#210, thanks @MaitreyeeDeshmukh)
  • Onboarding: in-app Product Tour button (#225, thanks @gaoharimran29-glitch)
  • MCP: get_ai_traffic (#148), get_prompt_volumes (#160), list_shopping_cards / get_product_visibility (#177), and prompt-level performance aggregation (#181) tools — each with a parallel REST endpoint (#148, #177, #181 thanks @Pallavikumarimdb)
  • Tests: Vitest infrastructure for both web/ (#202) and server/ (#249), plus unit tests for the CSV serializer (#219), classifyDomain / hostname helpers (#248), and parseResponse / countBrandMentions (#251) (all thanks @Pallavikumarimdb)
  • CI: lint + CI pipeline for the server/ package (#201, thanks @Pallavikumarimdb)
  • DX: seed now populates raw prompt_results.shopping_cards so the demo dashboard shows shopping data out of the box (#232)

Changed

  • Plans: server plan limits now read from the same source of truth as the web app, so cloud quotas stay in sync (#223)
  • Sidebar: tighter nav-item density (#166), removed the redundant Settings entry (#167), and moved the collapse toggle above the profile row with a restyle (#168)
  • Brands: brand list cards slimmed to a nav-menu shape (#154, #156), typography aligned with the Insights page (#179), softened active-card outline (#175), bolder breadcrumb avatar fallback (#174)
  • Agent: today's date is injected into the system prompt so time-window queries ("last 7 days") resolve correctly (#137)

Fixed

  • Brands: page no longer crashes — buttonVariants is now server-safe (#230)
  • Auth: the full reset-password flow is wired end-to-end (#151, #171)
  • Insights: show platform totals (#172, thanks @nanookclaw); group results by platform on both the insights and prompt-detail views (#235, #237, thanks @VrtxOmega); CSV export writes platform display names instead of raw slugs (#234); moved the raw results count out of the page header (#238)
  • Tracking: cloud snippet points at api.ansvisor.com (#218); Shopping sidebar entry is gated by the active brand instead of org-wide (#170, #173)
  • Team settings: show the role label instead of the raw enum value (#147, thanks @akagifreeez)
  • UI: ChatGPT avatar stays visible in light mode (#162, thanks @nanookclaw); PasswordInput merges caller className via cn (#212, thanks @MaitreyeeDeshmukh); icon-only buttons across the dashboard now have accessible names (a11y) (#253, thanks @BharadwajKanneveti)
  • Billing: removed a stray debug log from the Stripe checkout route (#184, thanks @krishnaprasharkp)
  • Self-host: Docker Compose image tags sync with the package version (#185, thanks @xianzuyang9-blip)

Docs

  • Added a Code of Conduct (Contributor Covenant) (#247), a backend server/ README (#188, thanks @titanniya542-spec), and fork instructions in CONTRIBUTING (#135, thanks @ayobamiseun)
  • Repo: GitHub issue forms + PR template (#233); README polish — Resources section, single H1 tagline, product-tour badge, banner image, www links, and marking the in-product AI assistant as shipped (#165, #197, #207, #214, #215, #216; thanks @beanscg, @n1dhiparate, @xzlknr)

Contributors

Huge thanks to everyone who contributed to this release: @Pallavikumarimdb, @MaitreyeeDeshmukh, @n1dhiparate, @nanookclaw, @VrtxOmega, @ayobamiseun, @akagifreeez, @beanscg, @xzlknr, @titanniya542-spec, @xianzuyang9-blip, @krishnaprasharkp, @gaoharimran29-glitch, and @BharadwajKanneveti. 🙌

[0.1.2] - 2026-05-31

Added

  • In-product AI agent: chat panel grounded in the MCP read tools, available on every cloud plan via BYOK — paste your own Anthropic API key in Settings → Agent. Self-host uses ANTHROPIC_API_KEY from env (#120, #121)
  • Settings → Agent: org-level Anthropic API key management for cloud customers; AES-256-GCM encrypted at rest, only last4 + saver metadata visible to org members, save/clear is admin-only (#121)
  • MCP: generate_content_brief tool that triggers the brief endpoint (#109)
  • MCP: update_opportunity_status tool for workflow transitions (#110)
  • MCP: get_competitor_comparison tool with share-of-voice (#116)
  • MCP: list_citations tool + REST endpoint (#117)
  • MCP: get_visibility_trend tool (visibility time-series) + REST endpoint (#118)

Changed

  • Insights: aggregate insights data in Postgres instead of pulling rows into Node — meaningful drops in p95 for orgs with large prompt-result tables (#114)
  • Repo: renamed from aeohub/ansvisor to ansvisor/ansvisor; all internal links + docs updated (#102)
  • Marketing: removed the in-app /pricing page; canonical pricing lives on ansvisor.com/pricing, and /pricing on the app redirects there (#119)
  • CI: ESLint now runs in CI alongside Prettier and TypeScript (#128, thanks @ayobamiseun); the 8 existing lint errors lurking in the codebase were cleared in the same window so the new check stays green (#133)

Fixed

  • Invite flow: clicks on invite emails now route through a new /auth/confirm route handler that does server-side verifyOtp and writes the session cookie before the user lands on the accept page. The previous flow ejected invitees to /sign-up, where Supabase's silent duplicate-signup obfuscation left them with no password set; the accept card now also asks for a password + full name before joining so the user can sign back in (#127, #129, #130)
  • Onboarding: align prompts to the selected plan's engine set on Stripe checkout success — Starter customers no longer see Growth-only platforms after upgrading via the onboarding flow (#111)
  • Billing: same alignment runs on every plan-change path (PATCH subscription, webhook, downgrade) so prompts stay consistent with the active plan regardless of which surface fired the change (#112)

[0.1.1] - 2026-05-26

Added

  • MCP server with API keys + list_brands and get_visibility_summary tools, exposed at /api/mcp (#20)
  • MCP: list_prompts / get_prompt and list_topics / get_topic tools, plus parallel REST endpoints (#35)
  • MCP: list_content_opportunities / get_content_opportunity tools + REST endpoints (#74)
  • Anthropic Skills: Ansvisor AEO Coach ships in two flavours — MCP tool for Claude Desktop / Code / Cursor / Zed, and standalone REST for claude.ai web (#23)
  • Analytics: PostHog + Vercel Analytics with self-host opt-in posture (#13)
  • Analytics: universal user identification and onboarding-funnel instrumentation (#30)
  • CSV export buttons on Topics (#53), Prompts (#54), and Answer Engine Insights (#73)
  • Citations: searchable prompt combobox filter (#55)
  • Sidebar: user profile chip (avatar + name) linking to settings (#52)
  • Prompts: Competition column with a 5-bar difficulty meter (#82)
  • Tracking: capture Perplexity shopping_cards into prompt_results (#83)
  • Tracking: capture Google AI Mode shoppingCards into prompt_results (#86)
  • Tracking: capture Microsoft Copilot shoppingCards into prompt_results (#87)
  • DX: supabase/seed.sql ships a populated local dashboard (one demo org, brand, prompts, ~120 prompt results, competitors, content opportunities, AI traffic logs) — [email protected] / demo123 (#75)
  • Tooling: Prettier configuration + CI workflow (format check & typecheck) (#80)

Changed

  • README: replaced the intro with a build-in-public manifesto (#90)
  • README / docs metadata: tagline updated to "AI Visibility & AI Search Optimization" (#89)
  • Docs: rewrote "What is Ansvisor?" around AI Search Visibility / GEO / AEO (#92)
  • README: stargazers CTA above "Why Ansvisor?" (#77)
  • Onboarding: signout button in the bottom-right corner (#68)
  • Settings: contact-us CTA opens the contact page (#81)
  • CI: auto-welcome first-time contributors on PRs only (#34, #59)

Fixed

  • Billing: block tracking + features for orgs without an active subscription (#56)
  • Citations: group raw model slugs under display names in the Platforms filter (#48)
  • Insights: adaptive Y-axis on the Brand vs Competitors chart (#37)
  • Insights: silence navigation-cancellation toast (#70)
  • MCP: use the app URL for the MCP endpoint (#33)
  • UI: ComboboxTrigger overflow — respect caller width and clip long values (#91)
  • Onboarding: preserve pending content opportunities (#63)
  • UI: sign-in / sign-up header logo points at the marketing site (#57)
  • UI: remove unused dashboard layout header (#36)
  • Refresh stale package-lock metadata (#51)

[0.1.0] - 2026-04-09

Added

  • Initial open-source release
  • Web frontend (Next.js 16) with dashboard, analytics, and content optimization
  • Backend server (Express) with multi-provider AI tracking (ChatGPT, Gemini, Perplexity, Grok, Claude)
  • Docker Compose setup for self-hosting
  • Multi-language support (13 languages, 18 regions)
  • Plan-based feature gating (self-hosted, starter, growth, enterprise)
  • Real-time brand visibility monitoring across AI search engines
  • Competitor tracking and content optimization suggestions
  • Prompt volume analysis
  • Stripe integration for cloud billing