ads_filter_options
ChatGPTThe valid filter values for the advertising tools — platforms, businesses, ad accounts, campaigns, ad sets and ads. Options cascade: pass what you already know and the response narrows to what is still reachable. Call this BEFORE filtering ads_report or ads_grid by an id, since ad-platform ids are assigned externally and cannot be guessed.
ads_grid
ChatGPTThe same advertising metrics as ads_report but as ranked, paginated rows grouped by one dimension. Use for 'which campaigns spent the most', 'how did each ad set perform', 'ad spend per day'. Set group_by to campaign (default), platform, business, account, adset, ad, ad_type, lead_source, currency or date. Sort with sort_by (spend default, or leads/clicks/impressions/cpl/cpc/cpm/ctr/name) and sort_dir. Page through with page; per_page maxes at 100. Accepts the same filters as ads_report plus search.
ads_report
ChatGPTPaid advertising performance combined across every connected platform (Meta, Google Ads, Custom Advertising): total spend, impressions, clicks, leads and the derived CPL/CPC/CPM/CTR, plus a per-platform split, a daily time series, and the top campaigns. Answers 'how much did we spend on ads?', 'what's our cost per lead?', 'which platform performs best?'. NOTE: this is PAID ADS — for the CRM's own email/WhatsApp campaigns use campaigns_list / campaigns_analytics instead. Spend is per-row currency and is NOT converted, so never sum across different currencies. Call ads_filter_options first if you need a platform or campaign id.
analytics_fields
ChatGPTWhich fields of ONE module can be grouped by, filtered on, bucketed over time, and summed. ALWAYS call this before analytics_group_by — field slugs differ per organization (a call's status is outgoing_call_status on some orgs and call_status on others) and custom modules are entirely org-specific, so guessing a slug fails. Returns group_by (every valid grouping, including the system columns record_owner / created_by / created_at), date_fields, numeric_fields (what you may sum or average) and aggregations.
analytics_group_by
ChatGPTBreak a module's records down BY a field and aggregate them — the tool for every 'X by Y' question: 'calls by user', 'deals by stage', 'contacts by source this month', 'total call duration per agent'. Returns groups as [{label, key, value}] with people and picklists already resolved to display names. Call analytics_fields first to get valid group_by / aggregation_field values. Owner groupings automatically span record_owner, the module's own owner field and created_by, so no record is missed or misattributed. Read-only.
analytics_metrics
ChatGPTThe catalog of shortcut analytics metrics, PLUS the list of modules you can aggregate directly. The shortcut list is small and is NOT the limit of what can be computed — any module in modules can be counted, summed or grouped via analytics_group_by. Rarely needed: prefer analytics_fields for a specific module.
analytics_run
ChatGPTRun a full report-builder config — the ESCAPE HATCH for shapes analytics_group_by cannot express: pivots (row x column cross-tabs), funnels, and multi-metric KPI cards. For an ordinary 'X by Y' breakdown use analytics_group_by instead; it is far harder to get wrong. Call analytics_fields first for valid slugs. Read-only: it computes but never changes data.
campaigns_analytics
ChatGPTPerformance analytics for one campaign (sends, delivered, opens, clicks, conversions, rates). Use for 'how did campaign X perform?'.
campaigns_get
ChatGPTFetch a single campaign by id, with its details.
campaigns_list
ChatGPTList marketing campaigns (newest first, paginated). Filter with search, status, channel. Use limit (1-100) and page.
dashboards_get
ChatGPTOne dashboard and its tiles. Each tile carries a runnable config — pass it to analytics_run to compute that tile's current numbers. Answers 'what's on my Sales dashboard' and 'what does it say now'.
dashboards_list
ChatGPTThe organization's saved dashboards. Use for 'what dashboards do we have', and to find the id for dashboards_get.
dashboards_meetings
ChatGPTThe asking person's meetings for a date (defaults to today). Answers 'what meetings do I have', 'am I free tomorrow'.
dashboards_productivity
ChatGPTThe current user's productivity summary — open tasks, tasks due today, today's meetings, and pinned notes. Use for 'what's on my plate today?', 'how productive is the team today?'.
dashboards_widget
ChatGPTOne productivity widget for the person asking: my_tasks, tasks_due_today, events_today, calendar, kanban or sticky_notes. Answers 'what's due today', 'what's on my plate'.
lists_get
ChatGPTFetch a single contact list by id.
lists_list
ChatGPTList static contact lists (paginated).
my_agenda
ChatGPTThe signed-in person's OWN plate: what is OVERDUE and what is DUE TODAY, across EVERY module at once (tasks, meetings, calls, deals, and any custom module with a due/scheduled date). This is the ONE tool for 'my overdue items', 'what's overdue', 'what's due today', 'what's on my plate', 'my to-dos', and ANY follow-up about the daily briefing ('what are the N overdue items you mentioned?'). USE THIS INSTEAD OF objects_search for anything overdue/due. It already knows, per module, which field is the due date, which is the status, and which status values mean done — so you must NOT guess a status field (there is no task_status; tasks use sales_task_status) or build your own 'past due and not completed' filter. It is also the SAME definition the daily briefing counts with, so its totals match the briefing's number exactly. Returns totals {overdue, today} (EXACT counts — use these for 'how many'), by_module (the same split per module), and records for ONE window (the fields that matter: label, module, due, owner, status). Pick the window with window ('overdue' is the default; 'today' for today's). The record list is paged: pass limit (1-100, default 50) and, for more, after = the previous response's paging.next_after. module narrows the list to one module. Every row is already scoped to this person and to what they may see — no owner filter needed.
notifications_list
ChatGPTThe asking person's in-app notifications and unread count. Answers 'anything I missed', and explains failures — quota warnings and limit alerts land here.
objects_get
ChatGPTFetch a single CRM record by its id, with all of its properties. Use this after objects_search / objects_list gives you an id, or when the user references a specific record id directly. Optionally look up by a unique field instead via id_property (e.g. 'email').
objects_list
ChatGPTList records of a CRM object type, newest first, with cursor pagination. Use after (from the previous response's paging.next.after) to fetch the next page, and limit (1-100, default 10) to bound the page size. Prefer objects_search when the user is filtering by a field value. sort is one key like 'created_at:desc'; properties is a comma-separated sparse fieldset.
objects_list_types
ChatGPTList every CRM object type (module) available here, including the custom modules this org built — each entry has slug (the value every other tool's module param takes), label, singular and plural. CALL THIS FIRST whenever a question names a module outside the eleven standard ones (contacts, companies, deals, products, quotes, invoices, payments, tasks, meetings, calls, notes). A custom module's slug is NOT its display name — 'Stores' may be stores, store or stores-1 — so match the user's word case-insensitively against label/singular/plural and use that entry's slug. Guessing a slug wastes a turn and returns a not-found you cannot distinguish from the module truly not existing. One call covers the whole conversation.
objects_search
ChatGPTSearch / filter records of a CRM object type. This is the tool for questions like 'find the contact named Prayas', 'open deals owned by X', 'contacts created this month', 'how many companies in India'. TWO WAYS TO SEARCH — pick the right one: 1. query — free text. Use this whenever the user names a person, company, email or phone and you do NOT know which field holds it ('find contact named Prayas' -> query='Prayas'). It matches a case-insensitive substring across every searchable field on the module at once, so you do not have to guess between first_name / last_name / full_name. Prefer it over guessing a field. 2. filterGroups — structured field filters, for conditions on a KNOWN property (stage, owner, amount, date). HubSpot-style DSL: a list of groups (OR between groups, AND within a group), each with filters of {propertyName, operator, value|values}. The two combine: passing both ANDs them (free text AND filters). VALID OPERATORS (exact strings — anything else is rejected with INVALID_FILTER): EQ, NEQ, LT, LTE, GT, GTE, BETWEEN, IN, NOT_IN, HAS_PROPERTY, NOT_HAS_PROPERTY, CONTAINS_TOKEN, NOT_CONTAINS_TOKEN, STARTS_WITH, ENDS_WITH. Note substring match is CONTAINS_TOKEN, not CONTAINS. Put the comparison in value for the single-value operators; use values for IN / NOT_IN (any length) and BETWEEN (exactly 2 — low, high); HAS_PROPERTY / NOT_HAS_PROPERTY take neither. Call properties_list first to get exact property names. Supports sorts ([{propertyName, direction: ASCENDING|DESCENDING}]), limit, and after. Max 5 groups, 6 filters/group. limit must be between 1 and 100 (never 0). To COUNT matching records, request limit: 100 and count the results; if the response includes paging.next, there are more than 100 — page with after and keep a running total, or tell the user it exceeds 100. Do not pass limit:0 to get a count — that is rejected.
plan_usage
ChatGPTThe plan's limits and how much of each is consumed — contacts, seats, storage, emails, API calls. Answers 'am I near my contact limit', 'how many seats are left', and explains why an action was blocked.
properties_list
ChatGPTList the field (property) definitions for a CRM module — each field's key/slug, label, and type. Use this to discover the exact property names to pass to objects_search filters or the properties/sort args of objects_list.
records_timeline
ChatGPTThe HISTORY of one record — what changed, when, and who did it. Every other tool returns a record's CURRENT state, so this is the only way to answer 'what happened on this lead?', 'who moved it to Won and when?', 'when did we last touch this account?'. Optional activity_type narrows to one kind of event.
reports_activities
ChatGPTActivity-type breakdown (calls, meetings, notes, emails, WhatsApp) for a date range, plus meetings-by-user. Use for 'team activity this month'. Same range params as reports_overview.
reports_overview
ChatGPTHigh-level sales/CRM overview KPIs for a date range (new contacts, connected, calls, meetings, emails, WhatsApp, lead-status distribution, daily activity trend). Use for 'how are things this week/month?'. Pass range for the window; it defaults to current_week, so omitting it on a question about another period returns the wrong numbers. The response echoes the window it actually used under range — check it before quoting figures.
reports_pipeline
ChatGPTSales pipeline breakdown — deal counts/amounts by stage, and deals by user — for a date range. Use for 'what's in the pipeline?', 'deals by stage'. Same range params as reports_overview.
reports_users
ChatGPTPer-user activity table for a date range — contacts, deals, calls and meetings per person. Answers 'who is my top performer', 'compare the team'. Attributes records by who CREATED them; for attribution by OWNER (which differs for inbound calls) use analytics_group_by. LIMITATION: this is a fixed table — the only filter is the date range. It CANNOT filter by a record's fields (lead source, stage, type, tag). When the question adds such a filter — 'contacts by user excluding lead source X' — do not refuse: rebuild the table with analytics_group_by, grouping by the user field and putting the filter in criteria. search matches a USER's name/email for narrowing the roster; it does not filter records.
saved_reports_list
ChatGPTThe reports this organization has already built and named. Prefer running one of these over reinventing an equivalent query — a saved report is the definition the business agreed on, and its numbers match what the customer sees on their own screen.
saved_reports_run
ChatGPTRun a saved report by id and return its numbers. Get the id from saved_reports_list. Read-only: the definition is loaded, executed and discarded.
search_everything
ChatGPTSearch across EVERY module and entity at once. Use when the user names something without saying what it is — 'find Acme', 'what do we have on john@x.com' — and you do not yet know whether it is a company, contact, deal or note. When the module IS known, objects_search is more precise. Minimum 2 characters.
segments_get
ChatGPTFetch a single segment by id, with its definition.
segments_kpis
ChatGPTKPI counters for one segment (total members, added/removed today, last processed). Use for 'how big is segment X?'.
segments_list
ChatGPTList audience segments (paginated). Filter with search, module_slug, type.
teams_get
ChatGPTFetch one team by id, with its members expanded to names.
teams_list
ChatGPTList teams with their leads and member counts. Use for 'what teams do we have', 'how big is the sales team'.
users_get
ChatGPTFetch one person by their user id. Use to resolve a single owner id to a name.
users_list
ChatGPTList the people in this CRM — names, ids, active state, team membership. Use it to turn an owner id from a record into a NAME, and to answer 'who works here', 'who is on the sales team', 'is X still active'. Records return ownership as ids, so without this any 'who owns...' answer is an unreadable hex string — never guess a name, look it up. Optional search matches name or email.
voice_quota
ChatGPTTelephony allowance: calls made this month, the cap, and the phone number limit. Answers 'how many calls can we still make'. For call RECORDS use objects_search on the 'calls' module, and for call breakdowns use analytics_group_by.
voice_routing
ChatGPTWhich phone number rings which agent, inbound and outbound. Answers 'who picks up calls to this number'. Requires the voice configuration permission.
webforms_list
ChatGPTThe organization's web forms. Use to find a form id, then webforms_stats or webforms_submissions.
webforms_stats
ChatGPTSubmission counts and conversion for one form. Answers 'how many leads came through the Contact Us form'. An approved submission becomes a record and loses its provenance, so this is the only way to see the form funnel.
webforms_submissions
ChatGPTThe submissions to one form, newest first, with their status (pending / approved / rejected) and captured data.
workflows_get
ChatGPTFetch a single workflow by id, with its definition.
workflows_list
ChatGPTList automation workflows (paginated). Filter with search, status, module_slug, trigger_type.
workflows_report
ChatGPTExecution report / stats for one workflow (sent, delivered, read, clicked, failed, rates). Use for 'how is workflow X performing?'. channel is 'whatsapp' (default) or 'email'; optional date range via date_from/date_to.
ads_filter_options
Claudeads_grid
Claudeads_report
Claudeanalytics_fields
Claudeanalytics_group_by
Claudeanalytics_metrics
Claudeanalytics_run
Claudecampaigns_analytics
Claudecampaigns_get
Claudecampaigns_list
Claudedashboards_get
Claudedashboards_list
Claudedashboards_meetings
Claudedashboards_productivity
Claudedashboards_widget
Claudelists_get
Claudelists_list
Claudenotifications_list
Claudeobjects_get
Claudeobjects_list
Claudeobjects_list_types
Claudeobjects_search
Claudeplan_usage
Claudeproperties_list
Clauderecords_timeline
Claudereports_activities
Claudereports_overview
Claudereports_pipeline
Claudereports_users
Claudesaved_reports_list
Claudesaved_reports_run
Claudesearch_everything
Claudesegments_get
Claudesegments_kpis
Claudesegments_list
Claudeteams_get
Claudeteams_list
Claudeusers_get
Claudeusers_list
Claudevoice_quota
Claudevoice_routing
Claudewebforms_list
Claudewebforms_stats
Claudewebforms_submissions
Claudeworkflows_get
Claudeworkflows_list
Claudeworkflows_report
Claude
