Get Google Search Console search analytics data.

Args: dimensions: Comma-separated dimensions (query, page, date, device, country). Default: "query" start_date: Start date YYYY-MM-DD (default: 31 days ago) end_date: End date YYYY-MM-DD (default: 3 days ago, due to GSC data lag) row_limit: Max rows to return (default: 500) country: 3-letter country code filter (e.g. usa, gbr, can). Empty for global.

Get aggregate Google Search Console totals (clicks, impressions, CTR, position).

Args: start_date: Start date YYYY-MM-DD (default: 31 days ago) end_date: End date YYYY-MM-DD (default: 3 days ago) country: 3-letter country code filter (e.g. usa, gbr, can). Empty for global.

Compare two date periods in Google Search Console side-by-side.

Args: period1_start: First period start YYYY-MM-DD period1_end: First period end YYYY-MM-DD period2_start: Second period start YYYY-MM-DD period2_end: Second period end YYYY-MM-DD dimensions: Comma-separated dimensions (default: "query") row_limit: Max rows per period (default: 50) country: 3-letter country code filter (e.g. usa, gbr, can). Empty for global.

Find which pages rank for a given query in GSC.

Args: query: Exact query string to filter on start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) country: 3-letter country code (e.g. "usa"); empty for global row_limit: Max pages (default: 50)

Bucket indexed pages by average position (1-3 / 4-10 / 11-20 / 21+).

Args: start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) country: 3-letter country code; empty for global

Find which queries drive impressions/clicks to a specific page.

Args: page_url: Full or partial page URL (uses GSC 'contains' if no protocol) start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) country: 3-letter country code; empty for global row_limit: Max queries (default: 100)

Pages whose position changed most between two periods (positive = improved).

Args: period_a_start: First (older) period start YYYY-MM-DD period_a_end: First period end YYYY-MM-DD period_b_start: Second (newer) period start YYYY-MM-DD period_b_end: Second period end YYYY-MM-DD country: 3-letter country code; empty for global top_n: How many top movers and losers to return each (default: 25)

Pages with high impressions but ~0 clicks (CTR opportunities).

Args: start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) min_impressions: Minimum impressions threshold (default: 200) country: 3-letter country code; empty for global row_limit: Max pages to inspect (default: 100)

Split GSC totals into branded vs unbranded query buckets.

Args: start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) country: 3-letter country code; empty for global brand_terms: Comma-separated brand regex terms (default: "acme")

Compare GSC totals across multiple countries.

Args: start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) countries: Comma-separated 3-letter country codes (default: "usa,gbr,ind")

Extract long-tail, conversational GSC queries — the closest proxy to what users type into LLMs.

Pulls all queries for the period, filters to those with >= min_words words, and returns them sorted by impressions. These 7+ word queries reflect natural-language problem statements rather than keyword searches, making them strong signals for GEO content strategy.

Args: start_date: YYYY-MM-DD (default: 31 days ago) end_date: YYYY-MM-DD (default: 3 days ago) min_words: Minimum word count to include (default: 7) country: 3-letter country code (default: usa) row_limit: Max queries to return after filtering (default: 100)

Get GA4 site engagement metrics (sessions, users, engagement rate, bounce rate).

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) country: Full country name filter (e.g. "United States", "United Kingdom") channel: Channel filter (e.g. "Organic Search", "Paid Search", "Direct")

Get GA4 traffic breakdown by channel (Organic Search, Paid Search, Direct, etc.).

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) country: Full country name filter (e.g. "United States")

Get top landing pages by sessions from GA4.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) limit: Max pages to return (default: 30) country: Full country name filter (e.g. "United States") channel: Channel filter (e.g. "Organic Search")

Run a custom GA4 report with any metrics and dimensions.

Args: metrics: Comma-separated metric names (e.g. "sessions,totalUsers,keyEvents") dimensions: Comma-separated dimension names (e.g. "date,country"). Optional. start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) country: Full country name filter channel: Channel filter limit: Max rows (default: 50)

Top landing pages broken out by sessionSource.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Full country name filter (e.g. "United States") channel: Channel filter (e.g. "Organic Search") limit: Max rows (default: 50)

Top landing pages by key events / conversions.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Full country name filter channel: Channel filter limit: Max rows (default: 50)

Compare two date periods in GA4 side-by-side over one dimension.

Args: period1_start: First period start YYYY-MM-DD period1_end: First period end YYYY-MM-DD period2_start: Second period start YYYY-MM-DD period2_end: Second period end YYYY-MM-DD dimension: Dimension to group by (default: "landingPagePlusQueryString") country: Country filter channel: Channel filter limit: Max rows in diff (default: 50)

Sessions / users / engagement rate broken out by country.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) channel: Channel filter (e.g. "Organic Search") limit: Max countries (default: 30)

Best-effort event timeline for a single user. Requires user-scoped clientId custom dimension exposed via GA4 — falls back to empty rows if not configured.

Args: client_id: GA4 clientId value to filter on start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) limit: Max events (default: 100)

Approximate funnel: count users who fired each event in order. Returns per-step user counts and step-to-step conversion rates.

Args: steps: Comma-separated event names in order (e.g. "page_view,form_start,form_submit") start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country filter channel: Channel filter

Split sessions/users into new vs returning.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country filter channel: Channel filter

Breakdown of a specific event by a chosen dimension (default: landing page).

Args: event_name: GA4 event name (e.g. "form_submit") start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country filter channel: Channel filter group_by: Dimension to break out (default: "landingPagePlusQueryString") limit: Max rows (default: 50)

Sessions / users / engagement broken out by device category.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country filter channel: Channel filter

Top referrers (sessionSource) excluding direct.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country filter limit: Max referrers (default: 30)

Aggregate GA4 traffic from LLM platforms (Perplexity, ChatGPT, Claude, Gemini, etc.).

Returns total LLM-referred sessions/users/key-events plus a per-source breakdown.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) country: Full country name filter (e.g. "United States") limit: Max source rows to return (default: 20)

Note: totals reflect only the rows returned (bounded by limit). With 8 known LLM sources and limit=20, this covers all sources in practice.

Get Google Ads campaign performance (cost, clicks, conversions, impression share).

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today)

Get Google Ads keyword performance with quality scores.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) limit: Max keywords to return (default: 30)

Get actual search terms triggering Google Ads.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today) limit: Max search terms to return (default: 50)

Search HubSpot deals with filters (READ-ONLY).

Args: filters: JSON string of HubSpot filter groups array. Example: '[{"filters":[{"propertyName":"dealstage","operator":"EQ","value":"closedwon"}]}]' properties: Comma-separated deal properties to return limit: Max deals to return (default: 20) sort_by: Property to sort by (default: "createdate") after: Cursor token from a previous call's next_cursor for pagination

Most recently CLOSED-WON deals, newest close date first (READ-ONLY).

Use this for ANY question about recent/last/latest closed-won deals. It resolves the correct deal-stage IDs automatically from the deal pipelines, so you never build a filter yourself. Won/lost status lives on the deal's dealstage (deal pipeline stage) — NEVER use lifecyclestage, which is a contact/company field and does NOT represent deal won status.

Args: limit: Number of deals to return (default: 5) properties: Comma-separated deal properties to return

Get a HubSpot company by ID (READ-ONLY).

Args: company_id: HubSpot company ID properties: Comma-separated properties to return

Get a HubSpot contact by ID (READ-ONLY).

Args: contact_id: HubSpot contact ID properties: Comma-separated properties to return

Get a HubSpot deal by ID (READ-ONLY).

Args: deal_id: HubSpot deal ID properties: Comma-separated properties to return

Get contacts associated with a HubSpot company (READ-ONLY).

Args: company_id: HubSpot company ID limit: Max contacts to return (default: 5)

Get notes and meetings for a HubSpot company (READ-ONLY).

Args: company_id: HubSpot company ID include_notes: Include notes (default: True) include_meetings: Include meetings (default: True) limit: Max items per type (default: 5)

Get page visit history for a HubSpot contact (READ-ONLY).

Args: contact_id: HubSpot contact ID limit: Max page visits to return (default: 50)

Search HubSpot contacts with filters (READ-ONLY).

Args: filters: JSON string of HubSpot filter groups array (default: "[]") properties: Comma-separated contact properties to return created_after: ISO date filter on createdate (>=) created_before: ISO date filter on createdate (<=) limit: Max contacts (default: 50) after: Cursor token from a previous call's next_cursor for pagination

Search HubSpot companies with filters (READ-ONLY).

Args: filters: JSON string of HubSpot filter groups array properties: Comma-separated company properties created_after: ISO date filter (>=) created_before: ISO date filter (<=) limit: Max companies (default: 50) after: Cursor token from a previous call's next_cursor for pagination

Get contacts and company associated with a HubSpot deal (READ-ONLY).

Get companies and deals associated with a HubSpot contact (READ-ONLY).

Get deals associated with a HubSpot company (READ-ONLY).

Args: company_id: HubSpot company ID properties: Comma-separated deal properties

Search HubSpot meetings within a date range, optionally by owner/outcome.

Args: start_date: ISO date >= (default: 7 days ago) end_date: ISO date <= (default: today) owner: HubSpot owner ID filter outcome: hs_meeting_outcome filter (e.g. "COMPLETED", "NO_SHOW") limit: Max meetings (default: 50) after: Cursor token from a previous call's next_cursor for pagination

Form submission events for a HubSpot contact (READ-ONLY).

Email opens/clicks/sent events for a HubSpot contact (READ-ONLY).

Meetings associated with a HubSpot contact (READ-ONLY).

Notes, meetings, and calls associated with a HubSpot deal (READ-ONLY).

Args: deal_id: HubSpot deal ID limit: Max items per activity type (default: 25)

Aggregate deals by a property over a date range (count + amount sum).

Args: start_date: ISO date >= createdate (default: 30 days ago) end_date: ISO date <= createdate (default: today) group_by: Deal property to group by (default: "dealstage") pipeline: Pipeline ID filter (optional) limit: Max deals to scan (default: 200)

% of recent records that have each property populated.

Args: object_type: "deals", "contacts", or "companies" (default: "deals") properties: Comma-separated properties to check sample_size: Recent records to sample (default: 200)

Value counts for a property over recent records.

Args: object_type: "deals", "contacts", or "companies" property_name: Property to tally sample_size: Recent records to sample (default: 500)

Identify what type of HubSpot object (deal, company, or contact) an ID belongs to.

Use this when a lookup returns 404 and you're unsure whether the ID is a deal ID, company ID, or contact ID.

Args: unknown_id: The HubSpot numeric ID to probe

Find contacts whose first/last analytics URL contains a pattern, plus their deals.

Args: url_pattern: URL substring (e.g. "/alternatives/chargebee") start_date: ISO date >= createdate (optional) end_date: ISO date <= createdate (optional) limit: Max contacts (default: 100)

Get top Bing search queries (US traffic) with clicks, impressions, CTR, position.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today)

Get top Bing pages by clicks (US traffic).

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today)

Compare two date periods in Bing side-by-side.

Args: period1_start: First period start YYYY-MM-DD period1_end: First period end YYYY-MM-DD period2_start: Second period start YYYY-MM-DD period2_end: Second period end YYYY-MM-DD dimension: "query" or "page" (default: "query") limit: Max rows in diff (default: 50)

Find which Bing pages rank for a given query.

Args: query: Exact query string start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) row_limit: Max pages (default: 50)

Find which Bing queries drive a specific page.

Args: page_url: Page URL start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) row_limit: Max queries (default: 100)

Get keywords Bing topically associates with example.com (impressions-ranked).

Unlike query stats (what users typed), these reflect Bing's own topical index signal — useful for understanding perceived authority and content gaps.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country code (default: "us") language: Language code (default: "en-US") limit: Max keywords to return (default: 100)

Get keywords Bing considers related to a seed keyword for this site.

Useful for content gap analysis and discovering adjacent topics.

Args: keyword: Seed keyword to expand start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) country: Country code (default: "us") language: Language code (default: "en-US")

Get Bing's weekly crawl statistics for the site.

Returns a time-series of crawl health: pages crawled, index count, and error counts (DNS, HTTP, timeout, robots-blocked) per week.

Get list of URLs with crawl issues reported by Bing.

Each entry includes the URL and decoded issue labels (e.g. not_found_404, robots_txt_blocked, timeout). Useful for actionable technical SEO fixes.

Split Bing query totals into branded vs unbranded buckets.

Args: start_date: YYYY-MM-DD (default: 28 days ago) end_date: YYYY-MM-DD (default: today) brand_terms: Comma-separated brand regex terms (default: "acme")

Adds a new named tab to an existing Google Doc and writes formatted content.

The service account must have Editor access to the target document.

doc_id: Google Doc ID (from the URL: /document/d//edit) tab_title: Title shown on the tab blocks: JSON array of content blocks. Supported types:

{"type": "h1", "text": "Title"} {"type": "h2", "text": "Section"} {"type": "h3", "text": "Subsection", "space_above": 18} {"type": "p", "text": "Body paragraph text"} {"type": "p", "text": "Bold label", "bold": true, "space_above": 8} {"type": "p", "text": "Evidence citation", "italic": true} {"type": "p", "text": "Strong", "color": [0.1, 0.55, 0.15]} {"type": "callout", "text": "Warning text", "bg_color": [1.0, 0.98, 0.8], "bold": true} {"type": "bullet", "text": "Bullet point text"} {"type": "spacer"} {"type": "image", "drive_file_id": "1abc...", "width": 450, "height": 280} {"type": "image", "uri": "https://.../img/", "width": 450, "height": 280} -- uri: use the URL returned by POST /img/upload on the MCP server {"type": "table", "headers": ["Col A", "Col B", "Col C"], "rows": [["val1", "val2", "val3"], ...], "header_color": [0.18, 0.42, 0.70]}

Flags for p blocks: bold (bool), italic (bool), color ([r,g,b] 0-1 floats), space_above (pt) Flags for h1/h2/h3: space_above (pt) callout bg_color presets: warning=[1.0,0.98,0.8], info=[0.88,0.94,1.0], success=[0.9,0.97,0.9]

Returns the tab_id and a link to the document.

Creates a new Google Doc owned by the service account and writes formatted content into its default tab.

title: Document title blocks: JSON array of content blocks (same format as docs_add_tab)

Returns the document URL.

GSC totals + top 10 queries + top 10 pages in one call.

Replaces calling gsc_totals, gsc_search_analytics(query), and gsc_search_analytics(page) separately. Use this as the starting point for any 'how is organic search doing?' question.

Args: start_date: Start date YYYY-MM-DD (default: 31 days ago) end_date: End date YYYY-MM-DD (default: 3 days ago) country: 3-letter country code (e.g. usa, gbr). Empty for global.

GA4 site engagement + channel breakdown + top 10 pages in one call.

Replaces calling ga4_site_engagement, ga4_channel_breakdown, and ga4_top_pages separately. Use this as the starting point for any 'how is our website traffic doing?' question.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today)

Full marketing overview: GSC organic + GA4 traffic in one call.

Use this as the first tool for any weekly or ad-hoc 'give me an overview of marketing performance' question. Returns GSC totals, top queries, top pages, GA4 engagement, and channel breakdown.

Args: start_date: Start date YYYY-MM-DD (default: 28 days ago) end_date: End date YYYY-MM-DD (default: today)

Complete, deterministic data pack for the weekly inbound demo report (READ-ONLY).

Does in Python everything the demo skill used to hand-orchestrate: computes the Mon–Sun target week, runs the exhaustive "Inbound - Organic" unclassified sweep (paginated to the end), fetches in-period demo-outcome deals, and fully enriches every deal (associations, company, contacts, notes, meetings, activity timeline, form submissions) with pre-computed attribution signals. The LLM's only remaining job is classification judgment + narrative.

Returns one JSON object: target/prev weeks, in_period_deals[], stale_unclassified[], converting_urls[], plus complete and warnings so partial pulls are explicit instead of silently improvised.

Args: week_ending: Anchor date YYYY-MM-DD; the target week is the most recent completed Mon–Sun on/before it (default: today). weeks_back: Completed weeks back to target (default: 1 = last week). max_contacts_per_deal: Cap on contacts enriched per deal (default: 3).

Complete, deterministic data pack for the weekly traffic report (READ-ONLY).

Folds the traffic skill's GSC / Bing / GA4 / Google Ads batches plus LLM-referral gathering into one Python call: computes the Mon–Sun target + previous weeks (GSC gets its ~3-day lag), pulls every metric, and pre-computes the cross-references the skill used to ask the LLM to derive — the "Bing leads Google" candidate list and any emerging AI-surface referrer not yet in the curated list.

Each platform batch is isolated so one API failure degrades to a warnings entry instead of aborting the pack.

Args: week_ending: Anchor date YYYY-MM-DD (default: today). weeks_back: Completed weeks back to target (default: 1 = last week).

List upcoming Acme events in a date range.

Args: start_date: Start date YYYY-MM-DD (default: today) end_date: End date YYYY-MM-DD (default: 30 days from today) year: Calendar year to query — "2025" or "2026" (default: "2026")

Search Acme events by name, type, location, or attendee.

Args: query: Text to search in event name or type (case-insensitive) event_type: Filter by type — e.g. "conference", "dinner", "webinar", "demo day" location: Filter by city or venue (case-insensitive) attendee: Filter by person attending — e.g. "saurabh", "apurv", "priyam" year: Calendar year — "2025" or "2026" (default: "2026")

Get full details for a specific event by name.

Args: event_name: Full or partial event name to look up (case-insensitive) year: Calendar year — "2025" or "2026" (default: "2026")

Get Acme events budget summary — total spend, breakdown by type and quarter.

Args: year: Calendar year — "2025" or "2026" (default: "2026") quarter: Filter to a quarter — "Q1", "Q2", "Q3", or "Q4" (optional) event_type: Filter by event type — e.g. "conference", "dinner" (optional)

Show all events a specific person is attending.

Args: person: Person's first name — e.g. "saurabh", "apurv", "priyam", "brandyn", or "arnab" start_date: Only show events from this date onward YYYY-MM-DD (default: today) end_date: Only show events up to this date YYYY-MM-DD (optional) year: Calendar year — "2025" or "2026" (default: "2026")

Fetch and read the linked event website for a specific event.

Use this when you need details not in the sheet — venue, agenda, speakers, registration info, pricing. Fetches the URL linked in the event name cell.

Args: event_name: Full or partial event name (case-insensitive) year: Calendar year — "2025" or "2026" (default: "2026")

Persist one week's classified demo report to Supabase (idempotent).

Upserts the week-summary row on week_ending and replaces that week's lead rows. Re-running a week converges — never duplicates. Returns a non-fatal warnings list if pipeline_usd does not reconcile with counted-inbound lead amounts (the week is still saved).

Args: week_summary_json: JSON object for one demo_report_weeks row. Required key: week_ending (YYYY-MM-DD, the Sunday). Other keys: week_start, total_inbound, unbranded, branded, direct, events, outbound_revivals, paid, flagged, ambiguous, ai_surface_leads, pipeline_usd, warnings, complete. leads_json: JSON array of per-lead rows. Each: deal_id, company, domain, classification, icp_fit, amount, dealstage, found_us, ai_surface_origin, hubspot_current_source, hubspot_update_to, rationale. week_ending is attached automatically.

Read all stored demo weeks for a month and return pre-aggregated totals.

Pure Supabase read — no datapack pull, no HubSpot, no re-classification. The monthly skill renders the returned dict to a Google Doc tab.

Args: month: Target month as YYYY-MM (e.g. "2026-05"). Returns weeks whose week_ending falls in that month, their leads, and summed totals. A month with no stored weeks returns an explicit note.

Return recent, classified, non-noise Reddit candidates to comment on.

Reads the monitor pipeline's reddit_hits + reddit_classifications tables (no live Reddit fetch). Candidates are ranked by bucket priority (lead_signal > competitor_mention > icp_discussion) then recency, and split into posts vs in-thread comments. Each candidate includes the permalink, subreddit, title/body, author, age, and classification context (bucket, persona, pain_points, sentiment, mentioned competitors).

Intended use: draft a genuinely helpful, native-voice comment for the ones worth engaging (open the permalink first to read existing replies so you don't repeat them), then call reddit_mark_engaged after posting.

Args: lookback_hours: Only candidates created within this window. Default 48. include_posts: Include top-level post candidates. Default true. include_comments: Include in-thread comment candidates (reply opportunities). Default true. limit: Max candidates returned after ranking. Default 25. exclude_engaged: Hide candidates already recorded via reddit_mark_engaged. Default true.

Record that a Reddit candidate was engaged so it stops re-surfacing.

Idempotent upsert on post_id — call it once a comment is actually posted.

Args: post_id: The candidate's post_id (e.g. "t3_abc123" for a post or "t1_def456" for an in-thread comment), exactly as returned by reddit_engagement_candidates. comment_url: Permalink of the comment that was posted (optional, for the audit trail). note: Free-text note, e.g. "skipped — off-topic" or "posted" (optional).