Skip to content

MCP Tools ​

All tools available on the MCP server, grouped by domain. Bold arguments are required. Every write tool also accepts dry_run and idempotency_key (see conventions).

A key only sees the tools whose scope it has. Scopes marked admin only (conversations:write, agents:run, agent_tools:execute) can only be granted by an account administrator.

Contacts ​

ToolScopeKindArguments
contacts_searchcontacts:readreadquery, email, phone, company, source, tag_ids, created_after, created_before, updated_after, updated_before, sort_by, sort_order, page, limit
contacts_getcontacts:readreadcontact_id
contacts_createcontacts:writewritename, email, phone, company, title, location, linkedin_profile_url, source, custom_fields, tag_ids
contacts_updatecontacts:writewritecontact_id, name, email, phone, company, title, location, linkedin_profile_url, custom_fields
contacts_add_tagscontacts:writewritecontact_id, tags, create_missing
contacts_remove_tagscontacts:writewritecontact_id, tags
contacts_add_notecontacts:writewritecontact_id, content

Opportunities ​

ToolScopeKindArguments
opportunities_searchopportunities:readreadquery, pipeline_id, stage_id, owner_user_id, contact_id, campaign_id, company, created_after, created_before, updated_after, updated_before, sort_by, sort_order, page, limit
opportunities_getopportunities:readreadopportunity_id, include_history
opportunities_createopportunities:writewritecontact_id, name, email, phone, company, title, pipeline_id, stage_id, value, score, notes, owner_user_id, sector_id, campaign_id, source, custom_fields, activate_agent_id, enroll_in_flow_id
opportunities_updateopportunities:writewriteopportunity_id, title, value, score, notes, owner_user_id, sector_id, expected_close_date, probability, custom_fields
opportunities_move_stageopportunities:writewriteopportunity_id, stage_id, stage_name, pipeline_name, discard_reason_id, notes

Conversations (inbox) ​

ToolScopeKindArguments
conversations_listconversations:readreadcontact_id, opportunity_id, status, created_after, created_before, updated_after, updated_before, last_message_after, last_message_before, page, limit
conversations_getconversations:readreadconversation_id, messages_limit
conversations_send_messageconversations:writewriteconversation_id, content, pause_ai
conversations_add_noteconversations:writewriteconversation_id, content
conversations_take_controlconversations:writewriteconversation_id, reason
conversations_transferconversations:writewriteconversation_id, sector_id, user_id

Tasks ​

ToolScopeKindArguments
tasks_searchtasks:readreadquery, status, assignee_id, contact_id, opportunity_id, playbook_run_id, task_pipeline_id, task_type, priority, overdue, due_after, due_before, updated_after, sort_by, sort_order, page, limit
tasks_gettasks:readreadtask_id, include_comments
tasks_referencetasks:readread
tasks_createtasks:writewritetitle, opportunity_id, contact_id, status, description, task_type, priority, due_date, end_date, assignee_ids, label_ids, task_pipeline_id, size, estimate_hours, reminder_minutes_before, recurrence
tasks_updatetasks:writewritetask_id, title, status, description, task_type, priority, due_date, end_date, assignee_ids, label_ids, task_pipeline_id, size, estimate_hours, reminder_minutes_before, recurrence
tasks_set_statustasks:writewritetask_id, status
tasks_add_commenttasks:writewritetask_id, content, mentions

Playbooks ​

ToolScopeKindArguments
playbooks_listplaybooks:readreadquery, include_inactive
playbooks_getplaybooks:readreadplaybook_id
playbook_runs_searchplaybooks:readreadquery, playbook_id, status, overdue, contact_id, opportunity_id, started_after, started_before, page, limit
playbook_runs_getplaybooks:readreadrun_id
playbooks_startplaybooks:runwriteplaybook_id, contact_id, opportunity_id, conversation_id, title, start_date, skip_weekends, timezone
playbook_runs_cancelplaybooks:runwrite (destructive)run_id, reason
playbook_runs_rescheduleplaybooks:runwriterun_id

Reference data ​

ToolScopeKindArguments
pipelines_listpipelines:readreadinclude_inactive
pipelines_getpipelines:readreadpipeline_id
tags_listtags:readread
users_listusers:readreadrole, is_active

Reports ​

ToolScopeKindArguments
reports_funnelreports:readreadpipeline_id, sector_id
reports_sales_summaryreports:readreadpipeline_id, sector_id, owner_user_id, from, to

AI employees ​

ToolScopeKindArguments
agents_listagents:readreadinclude_inactive
agents_chat_startagents:runwriteagent_id, lead_simulation
agents_chat_sendagents:runwritesession_id, message, event
agents_chat_endagents:runwrite (destructive)session_id
knowledge_searchknowledge:readreadquery, limit, type, agent_id, contact_id
agent_register_lead_sourceagent_tools:executewritesource, campaign, adset, creative, contact_id
agent_send_emailagent_tools:executewritesubject, body, contact_id, agent_id
agent_transfer_to_humanconversations:writewritereason, farewell_message, agent_id, conversation_id, contact_id
agent_schedule_meetingagent_tools:executewritemessage_prefix, agent_id, conversation_id, contact_id
agent_check_availabilityagent_tools:executereadskip_days, agent_id, conversation_id, contact_id
agent_book_meetingagent_tools:executewriteslot_id, title, agent_id, contact_id, conversation_id

Notes per tool ​

  • contacts_search / opportunities_search — query is a free-text partial match (name, email, company / title, contact). Combine updated_after with sort_by=updated_at for incremental sync.
  • contacts_create — needs name plus at least one of email, phone or linkedin_profile_url; a contact with the same email returns DUPLICATE_ERROR.
  • contacts_update — custom_fields replaces the whole object; tags are managed with contacts_add_tags / contacts_remove_tags.
  • contacts_add_tags — tag names, matched case-insensitively. Unknown names return TAG_NOT_FOUND unless create_missing: true (requires tags:write). Adding a tag fires the same automations as in the app (email flows, tracking).
  • opportunities_create — pass contact_id to link an existing contact, or name/email/phone/company to find-or-create one (matched by email). Without pipeline_id/stage_id the deal lands in the first stage of the account default pipeline. Without owner_user_id the account round-robin assigns an owner; the pipeline entry AI agent is activated as in the app.
  • opportunities_update — custom_fields are merged. Stage changes go through opportunities_move_stage.
  • opportunities_move_stage — by stage_id, or by stage_name (the opportunity's own pipeline is preferred; add pipeline_name to disambiguate, else AMBIGUOUS). Records history, sets won/lost timestamps and refreshes the Kanban in real time.
  • conversations_list — status is one of ai_active, manual, closed; use archived: true|false to filter archived conversations.
  • conversations_send_message — text only; sends through the conversation channel (WhatsApp, LinkedIn, Instagram, email, webchat). By default it pauses the AI employee on that conversation (pause_ai); ai_paused in the result says whether it actually did, with pause_reason otherwise (e.g. already_manual, no_agent).
  • conversations_take_control — status manual, AI paused, scheduled follow-ups cancelled, control_taken event recorded. Release control from the inbox.
  • conversations_transfer — user_id only assigns that user (sector unchanged); sector_id only moves to the sector and its round-robin picks the user (assigned_via: round_robin; queue_empty: true when nobody is in the queue); both moves and assigns that user without touching the round-robin. The user must be a member of the target sector, otherwise ASSIGNEE_NOT_IN_SECTOR. With dry_run the round-robin is only peeked. Same rules as Transfer Conversation. It does not pause the AI employee.
  • tasks_* / playbooks_* — the key sees what the user who created it sees in the app (admins see everything; other users their own work), and writes are authored by that user. Same rules as the Tasks and Playbooks endpoints.
  • tasks_update — assignee_ids and label_ids replace the current lists. The contact/opportunity of a task cannot change.
  • tasks_set_status — completing a recurring task creates the next occurrence (next_occurrence_id); completing a playbook step advances its run. Steps of a cancelled run cannot change.
  • playbooks_start — one task per step, due dates chained from start_date, assignees as configured in each step. With dry_run: true it returns the computed due date of every step without creating anything. Needs a contact, opportunity, conversation or title.
  • reports_* — same definitions as the in-app reports: active = not lost/discarded, won = active in a win stage.
  • AI-employee tools are described on the AI employees page.

Resources ​

URIScopeContent
getraze://accountany connected keyAccount id, name, type, enabled modules and the key's scopes
getraze://pipelinespipelines:readPipelines with ordered stages
getraze://tagstags:readTags with usage counts
getraze://usersusers:readTeam members
getraze://sectorssectors:readSectors (teams)
getraze://agentsagents:readFlow-based AI employees
getraze://tool-catalogagents:readThe native tools an AI employee can use, with JSON Schema

Prompts ​

PromptArgumentsWhat it does
qualify_leadcontact_id, pipeline_nameReads the contact, its conversations and opportunities, then proposes tags and a stage move — without executing them
daily_pipeline_reviewpipeline_id, owner_user_idFunnel + last 7 days + stale open deals, with three suggested actions

GetRaze - AI-Powered Lead Generation