Qobrix CRM MCP Server

Sum / avg / min / max / count of a numeric field across all matching records — optionally grouped by another field (top-N buckets). Paginates server-side (cap 20,000 rows) and computes in-process, which is how to get totals on Qobrix fields the API won't sort. Example uses: Total 2026 closed-sale volume: resource='contracts', field='final_selling_price_amount', op='sum', search='contract_type == "cos" and contract_status == "agreed" and date_of_contract >= "2026-01-01" and date_of_contract < "2027-01-01"'. Agent leaderboard by volume: same search, group_by='commission_to_2', top=5. Average list price by property type: resource='properties', field='list_selling_price_amount', op='avg', group_by='property_type', search='status == "available"'.

Invalidate cached Qobrix responses. Use this when CRM data was edited in Qobrix and you need fresh reads before the TTL (default 300s) expires. Without 'prefix': clears the whole cache (memory + Redis). With 'prefix': clears only matching keys, e.g. 'v1:request:opportunities' to refresh lead lists, or 'v1:request:properties' to refresh listings. Returns the number of cleared entries in each tier.

Inspect the response cache: hits/misses, in-memory size, in-flight coalesced requests, TTL (seconds), and Redis tier status. Use to verify the cache is paying off (high hits, low misses on repeated workflows) or to diagnose why a query feels slow (cold cache, Redis degraded). Stats are process-local; restart resets them.

Find contacts that appear on multiple deals or opportunities — the 'repeat customer' report every CRM analyst gets asked for. Three modes: 'buyers' (default) walks closed contracts → opportunity_id → opportunities.contact_name to identify the buyer behind each closed sale. 'sellers' walks contracts → property_id → properties.seller. 'leads' groups all opportunities by contact_name regardless of close. Returns each repeat contact with deal_count, total_volume, total_commission, first_deal, last_deal, and a deals[] breakdown. Example uses: 2026 repeat buyers: { kind: 'buyers', year: 2026, min_count: 2 }. All-time loyal customers: { kind: 'buyers', min_count: 3 }. Recurring sellers this year: { kind: 'sellers', year: 2026 }. Frequent enquirers (any close): { kind: 'leads', since_days: 365 }.

Count records matching a search expression — returns just the total count, no data payload. Faster and lighter than fetching records when you only need a number. Example uses: Total active sale listings: resource='properties', search='status == "available" and sale_rent == "for_sale"'. Q1 closed-won opportunities: resource='opportunities', search='status == "closed_won" and last_status_change >= "2026-01-01" and last_status_change < "2026-04-01"'. Contacts created this month: resource='contacts', search='created >= THIS_MONTH'.

Days-on-market analytics. Joins Contracts to Properties on property_id and computes the (close_date - listing_date) duration in days, then aggregates count / mean / median / p75 / p90 / min / max. Defaults: kind='sold' (cos + agreed), listing_date_field='listing_date' with per-row fallback to website_listing_date then created. Use group_by to break down by property_type, city, agent, or commission_to_2 (broker). Set include_outliers=true for the 5 longest and 5 shortest deals. Example uses: Overall DOM for 2026 sales: { kind: 'sold', year: 2026 }. DOM by property type: { kind: 'sold', year: 2026, group_by: 'property_type' }. DOM by city (with outliers): { year: 2026, group_by: 'city', include_outliers: true }. Reservation cycle: { kind: 'reserved', year: 2026 }.

Flexible 'deals' tool over the Contracts table — covers sales, rentals, listing agreements, and pipeline reservations in a single call. Defaults: kind='sale', contract_statuses=['agreed'] (so a no-arg call still means 'closed sale contracts'). Every default is overridable via kind / contract_types / contract_statuses / include_reserved / date_field / min_price / max_price / party filters / raw search. Sorts in-process across the matching set (for a single page of highest prices you can also use list/search with sort='-final_selling_price_amount') and resolves property_id, agent, commission_to, lawyers, bank, etc. to readable names. Example uses: Best 2026 closed sales: { year: 2026, top: 5 }. Best 2026 rentals: { kind: 'rental', year: 2026, top: 5 }. Under-contract pipeline value: { contract_statuses: ['reserved'], year: 2026 }. My deals: { assigned_to: 'CURRENT_USER', year: 2026 }. All revenue deals >€1M last 90 days: { kind: 'any_revenue', since_days: 90, min_price: 1000000 }. Agent leaderboard (by commission): { year: 2026, by: 'commission', top: 10 }.

Build a chronological timeline of edits to one field on one CRM record. Paginates GET /{resource}/{id}/changes and filters rows where the field appears in original or changed. Returns [{timestamp, user_id, user_name?, type, before, after}] sorted oldest-first. Use for: 'who changed source on this opportunity and when?'

Canonical sales funnel in one call. Six stages with conversion %: Leads (opportunities created in window) → Qualified (status in [open,won]) → Viewing (property-viewings created in window) → Offer (offers created in window) → Reserved (contracts with contract_status=reserved) → Closed (contracts cos + agreed). Each stage is scoped by the same date window and (optional) assigned_to / agent. Use stage_overrides to substitute a tenant-specific definition for any stage. Example uses: 2026 funnel: { year: 2026 }. My funnel last 90 days: { assigned_to: 'CURRENT_USER', since_days: 90 }. Agent Vera's funnel: { agent: '', year: 2026 }.

Get a single agent by UUID — resolve a RESO Member/ListAgent from a listing or lead. Returns { data: { id, ref, ... } }. Verified include: PrimaryContactContacts, User, Brands, AgencyAgents (parent brokerage). Agent 360° pattern: 1. Active listings: qobrix_search_properties with agent == '' and status == "available" 2. Active leads: qobrix_search_opportunities with agent == '' and status == "open" 3. Pipeline deals: qobrix_search_offers / qobrix_search_contracts linked to agent's opportunities 4. Activity: search Tasks, Calls, Meetings by the agent FK

Get a single call record by UUID — call detail in the Activity Tracking workflow. Returns { data: { id, ... } }. Include: AssignedToUsers, ContactContacts, RelatedOpportunityOpportunities, RelatedAgentAgents, CreatedByUsers.

Retrieve audit / change-log entries for a single CRM record (Qobrix LogAudit). Calls GET /api/v2/{resource}/{id}/changes. Each row has original (before), changed (after), user_id (who), timestamp, type (create/update/delete). IMPORTANT: LogAudit.source is the resource name (e.g. 'Opportunities'), NOT the lead marketing source field. Field-level before/after for e.g. lead source are keys inside original and changed objects. For a filtered timeline of one field use qobrix_field_change_history.

Get a single contact by UUID — the person/company detail view in the Lead-Contact Lifecycle. Returns { data: { id, first_name, last_name, email, phone, ... } }. Verified include: AssignedToUsers (RESO OwnerMember), User, Language, Organizations. To build the full contact journey (canonical 360° view): 1. Leads: qobrix_search_opportunities with contact_name == '' 2. Tasks: qobrix_search_tasks with contact == '' 3. Calls: qobrix_search_calls (search by contact FK) 4. Meetings: qobrix_search_meetings (search by contact FK) 5. Seller listings: qobrix_search_properties with seller == '' This contact 360° pattern is the core of Lead-Contact Lifecycle and Follow-up workflows.

Get a single contract by UUID — the deal close detail view in the Transaction Lifecycle. Returns { data: { id, ... } }. Verified include: Contacts (buyer/seller), PropertyIdProperties (the listing), OpportunityIdOpportunities (the lead), PaymentInstallments (payment schedule), ContractParties (all signatories), CreatedByUsers. Use include=['PaymentInstallments'] to audit the payment timeline.

Get a single email message by UUID — email detail in the Activity Tracking workflow. Returns { data: { id, ... } }. Include: Properties, ContactContacts, RelatedOpportunityOpportunities, Agent, Campaign.

Get dropdown/enum field options — discover valid values for status fields, types, and categories. Essential for building correct search expressions in any workflow. Returns { data: [...], pagination: { ... } }. Each item has: id, resource, field, label, value. Key examples for canonical workflows: Listing statuses: search='resource == "Properties" and field == "status"' → available/reserved/sold/withdrawn. Lead statuses: search='resource == "Opportunities" and field == "status"' → new/open/won/closed_lost. Property types: search='resource == "Properties" and field == "property_type"'. Task types: search='resource == "Tasks" and field == "task_type"'. Pair with qobrix_get_schema for complete field discovery.

Get properties linked to a lead — the 'demand-side matching' tool in the Sales Pipeline. Answers: 'What properties does this buyer want?' by returning properties attached to an opportunity. This is the canonical property suggestion pattern (RESO Demand/Solution matching). Reverse of qobrix_get_leads_by_property (which finds leads for a property). For free-language / ranked demand→supply (incl. projects), use qobrix_search_properties / qobrix_search_projects with search+boost derived from the lead's criteria. In the 8-stage buyer journey, use at the Solution/Viewing stage to prepare showing candidates. The id is the Opportunity UUID.

Get matching leads for a property — the 'supply-side matching' tool in the Sales Pipeline. Answers: 'Who is interested in this listing?' by returning leads whose criteria match the property. This is the canonical buyer-property matching pattern (RESO Prospecting). Use after listing a new property to find potential buyers in the existing lead database. Complement with qobrix_get_lead_properties for the reverse (demand-side) match, or qobrix_search_opportunities with boost for ranked matching (incl. projects). The propertyId is the Property UUID. Returns paginated results.

Get a single media item by UUID — media detail view in the Media Lifecycle. Returns media metadata including URL/path. Optionally specify size variant for images: 'thumbnail', 'medium', 'large'. Use 'thumbnail' for preview grids, 'large' for full-resolution display.

Get a single meeting by UUID — meeting detail in the Activity/Showing Lifecycle. Returns { data: { id, ... } }. Include: AssignedToUsers, ContactContacts, ViewingPropertyViewings (if it wraps a showing), CreatedByUsers.

Get a single offer by UUID — offer detail view in the Transaction Lifecycle. Returns { data: { id, ... } }. Include: OpportunityOpportunities (the lead/buyer intent), PropertyProperties (the listing), CreatedByUsers, ModifiedByUsers. Use to trace the deal chain: Offer → Opportunity → Contact + Property → Contract.

Get a single lead/opportunity by UUID — drill into a pipeline deal in the Sales Pipeline workflow. Returns { data: { id, ref, status, enquiry_type, contact_name, agent, ... } }. RESO mapping: status 'new' = Lead, 'open' = Active/SQL, 'won' = Closed Won, 'closed_lost' = Lost. Verified include: ContactNameContacts (the person), Properties (linked listings), AgentAgents (assigned broker, RESO ListAgent), PropertyTypes, ClosedLostReason, User, OwnerUsers. Canonical deal drill-down pattern: 1. Get the lead with include=['ContactNameContacts','Properties','AgentAgents'] 2. Check showing history: qobrix_list_viewings (search by related property) 3. Check offers: qobrix_search_offers for this lead 4. Check activity: search Tasks, Calls, Meetings linked to this opportunity FK: contact_name → Contacts, agent → Agents, owner → Users.

Get a single project by UUID — development detail view for off-plan/new-build listings. Returns { data: { id, ... } }. Verified include: Agents, Developer, Translations, LocationLocations, Assignee, Recommended, Favorites. To see project units: qobrix_search_properties with project == ''. To see project media: qobrix_list_media with related_model='Projects' and related_id=.

Get lat/lng coordinates for projects, useful for map display. Optionally filter with a search expression.

Get properties explicitly linked to a lead — the 'demand-side view' in the Sales Pipeline. Shows which properties a lead/opportunity has been matched to or expressed interest in. The id parameter is the Opportunity UUID (not the property UUID). In the canonical buyer journey: after Qualification, use this to review Solution/Viewing candidates. Alternative: qobrix_get_opportunity with include=['Properties'] for inline expansion. Complement with qobrix_get_leads_by_property for the reverse (supply-side) match.

Get a single property by UUID — the listing detail view in the Listing Lifecycle. Returns { data: { id, name, status, ... } }. RESO StandardStatus: 'available' = Active, 'reserved' = Pending, 'sold' = Closed, 'withdrawn' = Withdrawn. Use include[] to expand related entities inline: Verified: Agents, PropertyTypes, PropertySubtypes, PropertyViewings, Opportunities, Translations, SalespersonUsers, CreatedByUsers, LocationLocations, AgentAgents, SellerContacts, ProjectProjects. Workflow tips: include=['PropertyViewings'] → see showing history (Showing Lifecycle). include=['Opportunities'] → see interested leads (Sales Pipeline). include=['AgentAgents'] → listing agent (RESO ListAgent). include=['SellerContacts'] → seller/owner (RESO OwnerMember). FK fields: agent → Agents, seller → Contacts, project → Projects, salesperson → Users.

Get lat/lng coordinates for properties, useful for map display. Optionally filter with a search expression to get coordinates for a subset. Returns array of { id, coordinates } objects.

Get the full field schema for any CRM resource — the Schema Discovery tool. CALL THIS FIRST when unsure about field names for search, fields[], or sort. Returns every field with name, type, label, and validation rules (RESO Data Dictionary equivalent). Resources (case-sensitive): Properties, Contacts, Opportunities, Agents, Tasks, PropertyViewings, Projects, Offers, Contracts, Calls, Meetings, EmailMessages, Media, PropertyTypes, PropertySubtypes, PropertyFeatures, Locations. Canonical use: before building search expressions for any workflow (Listing, Sales Pipeline, Follow-up), call this to discover the correct field names. The schema is the source of truth. Pair with qobrix_get_field_options to see allowed enum values for dropdown fields. For search grammar, prefer qobrix_search_dsl_help.

Get a single task by UUID — task detail for pipeline management and follow-up audit. Returns { data: { id, ... } }. Include: AssignedToUsers (who), ContactContacts (for whom), Properties (which listing), RelatedOpportunityOpportunities (which deal), RelatedAgentAgents, TaskStatus, TaskTypes.

Get a single viewing by UUID — the showing detail view in the Showing Lifecycle. Returns { data: { id, ... } }. Verified include: PropertyViewingViewing. After retrieving, use the property and contact FKs to trace back to the listing and lead.

List agents (RESO Member resource) — real estate brokers managing listings and leads. Agents are the central actor across all workflows: they manage the Listing Lifecycle, drive the Sales Pipeline, and own Follow-up activities. Referenced by property.agent and opportunity.agent FKs. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Verified include: PrimaryContactContacts, User, Brands, AgencyAgents (parent agency). Workflow: to see an agent's full portfolio, search Properties and Opportunities by agent == ''.

List call records — part of the Activity Tracking / Follow-up workflow. Calls are touchpoints in the Lead-Contact Lifecycle; zero tolerance on missed follow-ups. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Include: ContactContacts (who was called), AssignedToUsers (who made the call), RelatedOpportunityOpportunities (which lead), RelatedAgentAgents, RelatedContractContracts. Workflow: use with qobrix_list_meetings and qobrix_list_email_messages for a complete activity timeline. To audit a contact's engagement: search calls, meetings, and emails by their contact UUID.

List contacts (RESO Contacts resource) from Qobrix CRM — the people/companies in the CRM. Core to the Lead-Contact Lifecycle: a Contact becomes a Lead when an Opportunity is created for them. RESO ContactType mapping: use opportunity.enquiry_type and property.seller FK to determine Buyer/Seller/Tenant role. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Key fields (42 total): first_name, last_name, name, email, phone, city, country, assigned_to (UUID → user, RESO OwnerMember), role, is_company, nationality, preferred_language, preferred_contact_method, consent fields, created, modified. Cross-references: property.seller → Contact UUID, opportunity.contact_name → Contact UUID, task.contact → Contact UUID, call/meeting.contact → Contact UUID. Workflow: to see a contact's full journey, search Opportunities, Tasks, Calls, and Meetings by their UUID.

List contracts (RESO TransactionManagement close) — the 'Deal Signing / Payment / Close' stage of the Sales Pipeline. A Contract finalizes the transaction: it links the Property, buyer/seller Contacts, and the Opportunity. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Verified include: Contacts, PropertyIdProperties, OpportunityIdOpportunities, PaymentInstallments (payment schedule), ContractParties (all signatories), CreatedByUsers. Canonical chain: Offer → Contract → Payment → Close. After contract signing, the property status moves to 'reserved' (Pending), then 'sold' (Closed).

List email messages — part of the Activity Tracking / Follow-up workflow and marketing campaigns. Emails are touchpoints in the Lead-Contact Lifecycle for nurturing and communication. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Include: ContactContacts (recipient/sender), Properties (listing references), Campaign (marketing campaign), RelatedOpportunityOpportunities, RelatedAgentAgents. Use with qobrix_list_calls and qobrix_list_meetings for a complete contact engagement timeline.

List media items (RESO Media resource) — photos, documents, floor plans attached to any entity. Core to the Media Lifecycle: every listing needs photos and floor plans; projects need brochures. NOTE: Returns { data: [...] } WITHOUT pagination metadata (unlike other endpoints). Must provide BOTH related_model AND related_id together, or omit both for global media. Listing media: related_model='Properties', related_id=''. Contact documents: related_model='Contacts', related_id=''. Project brochures: related_model='Projects', related_id=''. Alternative: qobrix_list_properties with media=true (default) embeds media in property responses. Each item has: id, media_type, reference_id, display_order, category, created.

List meetings — bridges the Activity Tracking and Showing Lifecycle workflows. Meetings can wrap property viewings (RESO Showing ↔ Meeting). Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Include: ContactContacts, ViewingPropertyViewings (linked showings), AssignedToUsers, RelatedOpportunityOpportunities, RelatedAgentAgents, CreatedByUsers, ModifiedByUsers. Workflow: include=['ViewingPropertyViewings'] to see which meetings are property showings vs general meetings. The first meeting/showing is the canonical trigger that moves a lead from follow-up to active sales.

List offers (RESO TransactionManagement) — the 'Decision Making' stage of the Sales Pipeline. An Offer formalizes a buyer's intent: it links an Opportunity (the lead) to a Property (the listing) with a price. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Include: OpportunityOpportunities (the lead), PropertyProperties (the listing), CreatedByUsers, ModifiedByUsers. Canonical transaction chain: Lead → Viewing → Offer → Contract → Close. After an offer is accepted, the next step is Contract creation (qobrix_list_contracts).

List opportunities/leads from Qobrix CRM — the central entity of the Sales Pipeline and Lead-Contact Lifecycle. In Qobrix, leads ARE opportunities. Each represents a buyer/renter/investor enquiry. RESO funnel mapping: 'new' = MQL/Lead, 'open' = SQL/Active, 'won' = Closed Won, 'closed_lost' = Lost. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. IMPORTANT: Avoid include=['Locations']. Safe verified includes: ContactNameContacts, Properties, AgentAgents, PropertyTypes, ClosedLostReason, User, OwnerUsers. Key fields (77 total): status, enquiry_type (buy/rent/invest), buy_rent, contact_name (UUID → Contact), agent (UUID → Agent), owner (UUID → user), source, area_of_interest, bedrooms_from/to, price ranges, next_follow_up_date, enquiry_date. Workflow patterns: Pipeline overview: group by status to see funnel distribution. Overdue follow-ups: status == "open" and next_follow_up_date <= NOW. New leads this week: created >= THIS_WEEK and status == "new". Two-way matching: demand→supply via qobrix_search_properties with lead criteria; supply→demand via this tool with boost against a listing (or native qobrix_get_leads_by_property).

List projects (property developments/complexes) — RESO Project equivalent for off-plan and new-build listings. Projects group multiple property units under one development, relevant in the Listing Lifecycle for new-builds. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Properties reference their project via the 'project' FK. Verified include: Agents, Developer, Translations, LocationLocations, Assignee, Recommended, Favorites. Workflow: to see units in a project, qobrix_search_properties with project == ''.

List properties (RESO Property resource) from Qobrix CRM. Core tool for the Listing Lifecycle — use to browse active inventory, track status changes, or audit listings. Returns { data: [...], pagination: { count, current_page, has_next_page, has_prev_page, page_count, limit } }. RESO StandardStatus mapping: 'available' = Active, 'reserved' = Pending/Under Contract, 'sold' = Closed, 'withdrawn' = Withdrawn. Key fields (184 total): name, ref, status, sale_rent (for_sale/for_rent), property_type, property_subtype, city, country, list_selling_price_amount, list_rental_price_amount, bedrooms, bathrooms, covered_area_amount, plot_area_amount, agent (UUID → qobrix_get_agent), seller (UUID → qobrix_get_contact), project (UUID → qobrix_get_project), salesperson (UUID → user). Workflow recipes: Active inventory → search: status == "available" and sale_rent == "for_sale". Recent listings → sort: '-created'. Buyer-property match → prefer qobrix_search_properties with search + boost for free-language demand. Listing media → follow up with qobrix_list_media(related_model='Properties'). PAYLOAD DEFAULTS: expand=false and media=false — FKs come back as UUIDs and media is not inlined (set expand=true / media=true explicitly when full nested objects or media URLs are needed). Prefer include[] for surgical expansion of specific associations. Default limit 10, max 100.

List tasks — the operational backbone of all canonical workflows: Follow-up, Pipeline Management, and Listing Checklists. Tasks drive daily agent cadence: overdue follow-ups, listing prep, closing steps, payment tracking. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Include: AssignedToUsers (task owner), ContactContacts (related person), Properties (listing), RelatedOpportunityOpportunities (lead), RelatedAgentAgents, RelatedContractContracts, TaskStatus, TaskTypes, CreatedByUsers, ModifiedByUsers. Canonical patterns: My open tasks: search with assigned_to == CURRENT_USER and status != "completed". Overdue: search with due_date <= NOW and status == "pending". Tasks for a lead: search with related opportunity FK. Listing checklist: search tasks linked to a property UUID.

List property viewings (RESO ShowingAppointment) — the Showing/Viewing Lifecycle tool. Viewings are the pivot between Follow-up and Active Sales: the first viewing marks a lead as actively engaged. Returns { data: [...], pagination: { count, current_page, has_next_page, ... } }. Each viewing links a property, a contact, and an agent/creator. Verified include: PropertyViewingViewing. Workflow patterns: This week's viewings: search with created >= THIS_WEEK. Property's showing history: qobrix_get_property with include=['PropertyViewings']. Meetings linked to viewings: qobrix_list_meetings with include=['ViewingPropertyViewings']. After a showing, the next pipeline step is Offer → Contract → Close.

Per-rep productivity roll-up. One call returns a wide row for a single user or a top-N leaderboard (default). Counts calls/meetings/emails/tasks/viewings/opportunities and deals closed; sums total_volume + total_commission from closed contracts. Bucketed by assigned_to across every resource. Example uses: Leaderboard 2026 by volume: { year: 2026 }. My scorecard last 30 days: { user: 'CURRENT_USER', since_days: 30 }. Top 5 by activities this quarter: { from: '2026-01-01', to: '2026-04-01', sort_by: 'activities', top: 5 }.

Relevance-ranked agent search (F1-optimized). TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences; limit/max_scan control top-N and pool size. With boost: _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'Agents'}) for DSL + fields. Example: boost=[{field:'ref',op:'contains',value:'LIM',weight:2}], limit=10, max_scan=100.

Search tenant-wide audit log for a CRM resource (Qobrix LogAudit). Calls GET /api/v2/{resource}/changes. Use to find who changed what across all records. Search examples: timestamp >= DAYS_AGO(30), user_id == "", primary_key == "". Pair with qobrix_top_field_changers for leaderboards or qobrix_field_change_history for one record.

Relevance-ranked contact search (F1-optimized). TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences scored over up to max_scan candidates; limit = top-N returned. With boost: rows get _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'Contacts'}) for DSL + fields. Examples: search='country == "CY"', boost=[{field:'city',op:'==',value:'Limassol',weight:2},{field:'is_company',op:'==',value:false,weight:1}], limit=10, max_scan=100. Hard-only: assigned_to == CURRENT_USER; name contains "Smith"; created >= THIS_MONTH.

Relevance-ranked contract search (F1-optimized). TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences; limit/max_scan for top-N and pool. With boost: _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'Contracts'}) for fields. Example: search='contract_type == "cos" and contract_status == "agreed"', boost=[{field:'final_selling_price_amount',op:'>=',value:300000,weight:2}], limit=10, max_scan=150. Prefer qobrix_deals for common closed-deal questions.

Return the full Qobrix Search Expression DSL reference so you can build correct search strings and boost[] clauses for ANY qobrix_search_* tool (properties, projects, contacts, agents, opportunities, viewings, tasks, offers, contracts). Includes operators, functions, date helpers, association paths, and the two-tier relevance recipe (hard search filter + soft boost ranking + limit/max_scan). Optional resource (e.g. 'Properties', 'Opportunities', 'Contacts') appends a field cheatsheet and live schema field names (schema fetch is response-cached). Call before composing free-language queries.

Relevance-ranked offer search (F1-optimized). TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences; limit/max_scan for top-N and pool. With boost: _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'Offers'}) for fields. Example: search='created >= THIS_MONTH', boost=[{field:'created',op:'>=',value:'2026-01-01',weight:1}], limit=10, max_scan=100.

Relevance-ranked lead/opportunity search (F1-optimized) — primary pipeline filter AND supply→demand matching. TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences scored over max_scan candidates; limit = top-N. With boost: _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'Opportunities'}) for DSL + buyer-criteria fields. Pipeline recipes: status == "open"; created >= THIS_WEEK and status == "new"; next_follow_up_date <= NOW and status == "open". Supply→demand (who wants this listing?): search='status in ["new","open"] and buy_rent == "buy"', boost=[{field:'area_of_interest',op:'contains',value:'Limassol',weight:3},{field:'bedrooms_from',op:'<=',value:3,weight:2},{field:'list_selling_price_to',op:'>=',value:400000,weight:2}], limit=15, max_scan=200. Works for projects too (native by-property does not).

Highly relevant project (development) search for free-language demand (F1-optimized). TWO-TIER RECIPE: (1) search = hard DSL must-haves (server filter → precision). (2) boost[] = soft weighted nice-to-haves scored over up to max_scan candidates (recall + ranking). (3) limit = how many ranked rows to return (default 10, max 100). With boost: top-N with _relevance and _matched; pagination.mode='ranked'. Without boost: fast path single cached page; pagination.mode='fast'. Call qobrix_search_dsl_help({resource:'Projects'}) for DSL + field cheatsheet. Example: search='city contains "Paphos"', boost=[{field:'starting_price_from',op:'<=',value:400000,weight:2},{field:'construction_stage',op:'==',value:'under_construction',weight:1}], limit=10, max_scan=150. When expand=true or media=true, max_scan is auto-capped at 100. If status='result_too_large' with _refine_required, ask the user to narrow then retry. All upstream pages are response-cached.

Highly relevant property search for free-language buyer demand (F1-optimized). TWO-TIER RECIPE: (1) search = hard DSL must-haves (server filter → precision). (2) boost[] = soft weighted nice-to-haves scored client-side over up to max_scan candidates (recall + ranking → better precision@top). (3) limit = how many ranked rows to return (default 10, max 100) — raise for more options, keep modest to avoid context overload. With boost: returns top-N with _relevance (score) and _matched (which boosts hit); pagination.mode='ranked' and pagination.scanned shows pool size. Without boost: fast path — single cached list page (pagination.mode='fast'). Call qobrix_search_dsl_help({resource:'Properties'}) before composing queries. Examples: Hard only: search='status == "available" and sale_rent == "for_sale" and city contains "Limassol"'. Demand match: search='status == "available" and sale_rent == "for_sale"', boost=[{field:'sea_view',op:'==',value:true,weight:3},{field:'bedrooms',op:'>=',value:3,weight:2},{field:'list_selling_price_amount',op:'in',value:'200000..600000',weight:2}], limit=15, max_scan=200. PAYLOAD: keep expand=false / media=false for search; when either is true, max_scan is auto-capped at 100. If a result returns status='result_too_large' with _refine_required, ask the user to narrow the query (filters, fields[], smaller limit, drop expand/media) then retry. All upstream pages are response-cached (QOBRIX_CACHE_TTL, default 300s).

Relevance-ranked task search (F1-optimized) — daily ops for pipeline and follow-up. TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences; limit/max_scan for top-N and pool. With boost: _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'Tasks'}) for fields. Recipes: search='assigned_to == CURRENT_USER and status != "completed"', boost=[{field:'due_date',op:'<=',value:'2026-12-31',weight:3}], limit=15, max_scan=100. Overdue hard filter: due_date <= NOW and status == "pending".

Relevance-ranked viewing search (F1-optimized). TWO-TIER: search = hard must-haves; boost[] = soft weighted preferences; limit/max_scan for top-N and pool. With boost: _relevance + _matched; pagination.mode='ranked'. Call qobrix_search_dsl_help({resource:'PropertyViewings'}) for fields. Example: search='created >= THIS_WEEK', boost=[{field:'created',op:'>=',value:'2026-01-01',weight:1}], limit=10.

Start interactive Qobrix sign-in (Mode C only). When not connected, returns a Sign In to Qobrix link (or native URL elicitation) for the user to complete login + 2FA + consent. When already connected, reports the current identity. In Mode A/B this is a no-op — credentials come from env / request headers. Mode C uses a per-user encrypted session vault keyed by the chat identity (X-Chat-Platform / X-Chat-User-Id). Deliver the Sign In link only to that individual — never post it into a shared/group thread.

Sign out of Qobrix (Mode C only). Fully revokes the current user's session: calls the Authorization Server /disconnect (deletes the minted Qobrix API key and clears AS tokens/vault), then clears this user's local encrypted session vault. Other users' vaults on this MCP process are not affected. In Mode A/B there is no interactive session to clear.

Find live opportunities (default status in [new,open]) that have no recent activity. An opportunity is 'stale' when no call/meeting/email/task touched it within since_days AND the opportunity itself wasn't modified within that window. Sorted oldest-modified first. Example uses: Silent leads (default, 30d): {}. Aggressive cadence: { since_days: 7 }. My silent leads: { assigned_to: 'CURRENT_USER' }.

Time-bucketed metric (count / sum / avg / min / max) over any resource. Paginates once over the window, buckets in-process by day/week/month/quarter/year (default 'month'). Per-resource default date_field: contracts→date_of_contract, opportunities/properties/calls/etc.→created. Use search to filter (e.g. only closed sales). Set compare_to_prior=true to also fetch the prior identical-length window for YoY %. Example uses: Monthly 2026 closed-sale volume with YoY: resource='contracts', metric='sum', field='final_selling_price_amount', year=2026, search='contract_type == "cos" and contract_status == "agreed"', compare_to_prior=true. Weekly lead intake last 90 days: resource='opportunities', bucket='week', metric='count', since_days=90. Quarterly listing additions 2026: resource='properties', bucket='quarter', year=2026, metric='count'.

Leaderboard: which users most often edited a given field (tenant-wide audit scan). Paginates GET /{resource}/changes, counts rows where field appears in changed, groups by user_id. Returns [{user_id, user_name?, change_count, latest_change_at, sample_record_ids}]. Use for: 'who rewrites lead source to direct most often in the last 6 months?'

Top-N rows of any resource sorted by a numeric or ISO-date field. Paginates server-side (cap 20,000 rows), sorts in-process, and resolves common FK fields (property_id, agent, owner, assigned_to, commission_to, commission_to_2, contact_name, salesperson, seller, project, developer_id, campaign_id) into readable names so the agent doesn't have to chain lookups. Use this for full-dataset top-N, or when list/search sort returns no rows on a nullable/computed field (e.g. opportunities.budget). For a single page of highest list prices, prefer qobrix_list_properties / qobrix_search_properties with sort='-list_selling_price_amount'. Example uses: Top 2026 closed sales: resource='contracts', sort_by='final_selling_price_amount', search='contract_type == "cos" and contract_status == "agreed" and date_of_contract >= "2026-01-01" and date_of_contract < "2027-01-01"'. Largest active listings (full scan): resource='properties', sort_by='list_selling_price_amount', search='status == "available" and sale_rent == "for_sale"'. Most recently modified leads: resource='opportunities', sort_by='modified'.

Aggregate records by a field and return the top N most-frequent values with counts. This tool paginates through all matching records server-side — no client-side scripting needed. Caps at 20,000 records for safety; use a search filter to narrow large datasets. Example uses: Top listing developers: resource='properties', field='developer_id', search='status == "available" and sale_rent == "for_sale"', resolve=true. Lead sources: resource='opportunities', field='source', search='created >= "2026-01-01"'. Agent workload: resource='opportunities', field='owner', search='status == "open"', resolve=true. Property type mix: resource='properties', field='property_type', search='status == "available"'. Set resolve=true to convert UUIDs (developer_id, agent, owner, etc.) to human-readable names.

Return the current Qobrix user profile, capabilities, and portals (GET /api/v2/session/). In Mode C with no session for this chat identity, surfaces a Sign In link so the user can authenticate first. Also includes the OAuth subject when available. Use to confirm which CRM identity the agent is acting as for this user.

Win-rate analytics over opportunities. Globally returns counts of new / open / won / closed_lost plus win_rate_pct = won / (won + closed_lost). Optional group_by slices by source, enquiry_type, owner, agent, or closed_lost_reason_id (multi-dim arrays supported). Window applied to last_status_change (fallback modified). Set include_top_losses=true to also receive the 10 most-recent closed_lost opportunities with reason labels and details. Example uses: Overall 2026 win rate: { year: 2026 }. Win rate by lead source: { year: 2026, group_by: 'source' }. Top loss reasons last 90 days: { since_days: 90, group_by: 'closed_lost_reason_id' }. My win rate: { assigned_to: 'CURRENT_USER', year: 2026 }. Forensic loss review: { year: 2026, include_top_losses: true }.