MCP App Store
by KeenEthics
monday CRM icon

monday CRM MCP Server for Claude and ChatGPT

Overview

Run your monday CRM in plain language. Build a pipeline from scratch, start the day with a ranked deal briefing, spin up a forecast dashboard, audit board health, clean up messy data in bulk, and turn meeting notes into deal updates. Every skill writes back into monday as a real update, doc, or dashboard. Built on the official monday MCP connector.

How to connect

  1. Claude: Open the monday CRM plugin listing for Claude and install it in Claude Code. Claude listing
  2. ChatGPT: Open the monday CRM listing in the ChatGPT app directory and choose Connect. ChatGPT listing monday CRM exposes 134 tools here.

Tools

activate-sequence

ChatGPT
Activate a sequence so it can enroll contacts and run its steps. Activate a sequence (sets status to ACTIVE, enabling enrollment and step execution). Convenience alias for update-sequence-status when direction is known to be ACTIVE.

agent_catalog

ChatGPT
Browse the account-wide catalog of available trigger types and skills for monday platform agents. READ-ONLY — no agent_id required. Use this tool to discover what's available BEFORE wiring anything to a specific agent. ACTIONS: - list_triggers: { block_reference_ids? } — returns available trigger types. Each entry has block_reference_id (required for manage_agent_triggers action:"add"), name, description, field_schemas (describes field_values shape), and required_fields (fields to collect from the user). Note: only triggers that can be added programmatically appear here. OAuth/3rd-party triggers (Slack, Gmail, Salesforce, etc.) require user setup in the monday.com UI and will not appear here. - list_skills: {} — returns available skills with id, name, description. Never guess or invent a skill id — always look it up here before calling manage_agent_skills action:"add". USAGE EXAMPLES: - List all trigger types: { "action": "list_triggers" } - Fetch specific trigger: { "action": "list_triggers", "block_reference_ids": ["some-block-ref-id"] } - List all skills: { "action": "list_skills" } RELATED TOOLS: - manage_agent_jobs — use block_reference_id from list_triggers to configure jobs with nested triggers - manage_agent_triggers — use block_reference_id from list_triggers to attach a trigger to a specific agent - manage_agent_skills — use skill id from list_skills, or action:"create" to author a new skill, then attach to an agent - manage_agent — manage the agent entity itself (create, update, delete, activate, etc.)

all_api_read

ChatGPT
Execute read-only GraphQL queries against the monday.com API. Only queries are accepted — mutations are rejected with an error before the request is sent. Use the get_type_details tool first to understand the schema before crafting your query.

all_api_write

ChatGPT
Execute GraphQL mutations against the monday.com API to create, update, or delete data. Only mutations are accepted — queries are rejected with an error before the request is sent. Use get_graphql_schema and get_type_details tools first to understand the schema before crafting your mutation.

all_monday_api

ChatGPT
Execute any monday.com API operation by generating GraphQL queries and mutations dynamically. Make sure you ask only for the fields you need and nothing more. When providing the query/mutation - use get_graphql_schema and get_type_details tools first to understand the schema before crafting your query.

all_widgets_schema

ChatGPT
Fetch complete JSON Schema 7 definitions for all available widget types in monday.com. This tool is essential before creating widgets as it provides: - Complete schema definitions for all supported widgets - Required and optional fields for each widget type - Data type specifications and validation rules - Detailed descriptions of widget capabilities Use this tool when you need to: - Understand widget configuration requirements before creating widgets - Validate widget settings against official schemas - Plan widget implementations with proper data structures The response includes JSON Schema 7 definitions that describe exactly what settings each widget type accepts.

board_insights

ChatGPT
This tool allows you to calculate insights about board's data by filtering, grouping and aggregating columns. For example, you can get the total number of items in a board, the number of items in each status, the number of items in each column, etc. Use this tool when you need to get a summary of the board's data, for example, you want to know the total number of items in a board, the number of items in each status, the number of items in each column, etc.[REQUIRED PRECONDITION]: Before using this tool, if new columns were added to the board or if you are not familiar with the board's structure (column IDs, column types, status labels, etc.), first use get_board_info with filters.columns.only to get column metadata without fetching views. This is essential for constructing proper filters and knowing which columns are available.[IMPORTANT]: For some columns, human-friendly label is returned inside 'LABEL_<column_id' field. E.g. for column with id 'status_123' the label is returned inside 'LABEL_status_123' field.

build-crm-setup

ChatGPT
Build approved monday CRM boards from a propose-crm-setup proposal. Create the CRM boards from the most recent propose-crm-setup proposal. Requires its setup_token and explicit user confirmation. Never call this in the same turn as propose-crm-setup: the user must first see the proposal and approve it in a later message, and the original request to set up a CRM is not approval. Renaming a board repoints its board relations at the renamed board, and reusing existing boards still creates the remaining proposed ones. Results that plan or create boards return a newer setup_token; a later build-crm-setup retry passes that newest token. Before writing, it checks the final names again and returns COLLISION without creating anything for unresolved conflicts. It returns PROPOSAL_OUTDATED, also without creating anything, when the approved proposal can no longer be reproduced, and PLAN_EXPIRED when its token expired; in both cases propose again with the same setup_token. It builds each board through a template or the monday board generator, verifies the result, and adds only fields required for a usable CRM. Confirmation and build results carry warnings the user needs to hear. Once boards exist, the result suggests connecting the user's email, Gmail or Outlook by detected provider (unless a CRM email account is already connected), and a calendar (unless one is already connected) through show_connect_integration; it also tells you to ask, each as its own question, whether to import Gmail contacts into the Contacts board and whether to add upcoming calendar meetings to the Deals board. It never imports data itself, and never deletes or overwrites an existing board. Scope: this tool builds CRM structure only. It does not import records from another CRM, build pure project-delivery workflows, model nested client-of-client accounts, split one motion into product or regional pipelines, enforce per-assignee privacy, or undo created boards. Saved views, automations, and mirror columns are not created by this version. Phone-first and field-sales teams can use the boards but should skip the optional email connection. Email and calendar consent must be completed by the user in a browser. Importing contacts from Gmail into the Contacts board is optional, offered after the boards are built, and happens only if the user agrees to it.

change_item_column_values

ChatGPT
Change the column values of a single item on a monday.com board. [IMPORTANT] If you need to update multiple items in one call, use update_items instead of calling this tool in a loop. Otherwise: change the column values of a single item in a monday.com board. [REQUIRED PRECONDITION]: Before using this tool, if new columns were added to the board or if you are not familiar with the board's structure (column IDs, column types, status labels, etc.), first use get_board_info with filters.columns.only to get column metadata without fetching views. This is essential for constructing valid column values. For board-relation linking tasks, call link_board_items_workflow before using this tool. Board-relation and dependency responses include linked item IDs up to 1000. Larger lists are omitted with a count and a message confirming the update succeeded.

check-integration-connection-status

ChatGPT
Check whether an in-progress integration connection has finished connecting. Report whether an OAuth flow started by show_connect_integration has completed yet, and finish any product-side setup the integration needs (for gmail / outlook, registering the mailbox as a CRM sending account, plus importing contacts for gmail with importData; for calendar with importData, adding upcoming external meetings as contacts linked to boardId). Not for querying already-connected accounts — use get-connected-email-accounts for that.

connect_external_agent

ChatGPT
Connect a custom external agent (an agent running on your own server/infra) to monday.com. { custom: { name, callback_url? } } Returns the new agent_id plus a one-time signing_secret and api_token used to verify webhook requests and call the monday.com API/MCP server — both are shown ONLY in this response, so capture them immediately. RULES: - Omitting callback_url creates the agent without a webhook — it won't be mentionable/assignable until one is added. - This tool is for CUSTOM agents only. For Claude, OpenAI, and other supported providers, use manage_agent.

create-sequence

ChatGPT
Create a new outreach sequence on a board. Create a new sequence on a board. Always created INACTIVE; use activate-sequence to enable it before the sequence can run. Call this tool directly with whatever arguments you have — do not pre-fetch boards or email accounts first. Never set ordinals or UUIDs. Never fabricate steps.

create-timeline-item

ChatGPT
Create a new structured activity entry (call, meeting, email) on a CRM item's timeline. Create a new structured activity entry on a CRM item's timeline. Returns the created timeline_item_id. For freeform notes (meeting outcomes, follow-ups), use create-timeline-note instead.

create-timeline-note

ChatGPT
Add an unstructured note (meeting outcome, follow-up reminder, or rep context) to a CRM item's timeline. Log a freeform text note on a CRM item's timeline — meeting outcomes, follow-up reminders, context for the next rep, or any unstructured observation.

create_action

ChatGPT
Use to save a reusable action (a stored code script) for later execution by run_action. Save a reusable action (a stored code script). Variables are injected as environment variables (access via os.environ in Python, process.env in JS/TS). Recommended: Test your code with execute_code before saving to ensure it works correctly. Network access is restricted to the following hosts: [api.monday.com/, mcp.monday.com/mcp]. Requests to any other host will be blocked. Example: name: "Get board items", description: "Fetches items from a board", language: "python", code: "import requests\nprint('done')"

create_agent

ChatGPT
Create a monday platform agent and configure it in one call: the agent, its jobs and triggers, tools, knowledge, and skills. Use this instead of manage_agent action:"create" for every new agent. Look up every id first, then call this once: - trigger block_reference_id and field_values: agent_catalog action:"list_triggers" - tool block_ids and built_in keys: manage_agent_tools action:"search" - skill ids: agent_catalog action:"list_skills" - board and doc ids for knowledge: the user's boards and docs The agent is created INACTIVE. If creating the agent itself fails, nothing is created. Each later step runs even when an earlier one fails, and the result lists every step with its outcome. If a step failed, keep the agent and retry only that step with manage_agent_jobs, manage_agent_triggers, manage_agent_tools or manage_agent_knowledge on the returned agent id. Never call this tool again for the same agent. Tools come back as added, needs_connection, needs_activation or skipped, exactly like manage_agent_tools action:"add". Only tools under added are ready. USAGE EXAMPLE: { "prompt": "Every Monday, review the Projects board (id 123) and post a summary of slipping projects as an update.", "jobs": [{ "name": "Weekly slip report", "instructions": "Read board 123, find items past due...", "triggers": [{ "block_reference_id": "10380", "field_values": { ... } }] }], "tools": [{ "block_ids": [456] }, { "built_in": ["web_search"] }], "knowledge": [{ "resource_id": "123", "scope_type": "BOARD", "permission_type": "READ" }] } RELATED TOOLS: - manage_agent — read, update, activate or delete the agent afterwards

create_automation

ChatGPT
Creates an automation on a monday board from a structured natural-language description. Use this tool only when you know: - boardId - the user's intended trigger - at least one intended action - any details the user provided that are relevant to the trigger, conditions, or actions The caller does not need to know the exact available automation blocks or their required fields. Describe the user's intent clearly — the tool will translate that intent into supported blocks and values. If a required detail is missing from the user's request, ask for clarification before calling the tool. If the tool returns status: "needs_clarification", present the unresolved fields to the user, gather answers, then call the tool again. Describe the automation in this format: Trigger: When <the event that should start the automation> Details: <relevant detail>: <value> Conditions: - Only if <condition that should be true> Details: <relevant detail>: <value> Actions: - <action the automation should perform>: <relevant detail>: <value> Rules: - Use one trigger. - Conditions are optional. - Multiple conditions mean AND. - Use one or more actions. - Do not use branching. - Use natural language, not block IDs or internal field names. - Actions may reference values from the trigger context, such as "{{item name}}", "{{creator}}", "{{status}}", "{{group}}", or "{{board}}". Terminology: - Trigger: the event that starts the automation, such as "when a new item is created". - Conditions: optional requirements that must be true before actions run. - Actions: what the automation does when it runs. Example: Trigger: When a new item is created Actions: - Send a notification: Recipient: John Snow Title: Important Update Message: The item "{{item name}}" was created. - Move the item to a group: Group: Top group

create_board

ChatGPT
Create a monday.com board. Use creationPrompt to describe how you want the board to be built

create_column

ChatGPT
Create a new column in a monday.com board. [REQUIRED PRECONDITION]: If the column needs type-specific configuration (columnSettings) — e.g. status/dropdown labels, formula definitions, number units — first call get_column_type_info with fetchMode "schema" for that column type to learn the valid settings structure. Do not guess the settings shape. To give the new column AI behavior, create it here first, then call configure_ai_column.

create_dashboard

ChatGPT
Use this tool to create a new monday.com dashboard that aggregates data from one or more boards. Dashboards provide visual representations of board data through widgets and charts. Use this tool when users want to: - Create a dashboard to visualize board data - Aggregate information from multiple boards - Set up a data visualization container for widgets

create_doc

ChatGPT
Create a new monday.com doc either inside a workspace or attached to an item (via a doc column). After creation, the provided markdown will be appended to the document. LOCATION TYPES: - workspace: Creates a document in a workspace (requires workspace_id, optional doc_kind, optional folder_id, optional docOwnerIds) - item: Creates a document attached to an item (requires item_id, optional column_id, optional docOwnerIds) USAGE EXAMPLES: - Workspace doc: { location: "workspace", workspace_id: 123, doc_name: "My Doc", doc_kind: "private" , markdown: "..." } - Workspace doc in folder: { location: "workspace", workspace_id: 123, doc_name: "My Doc", folder_id: 17264196 , markdown: "..." } - Item doc: { location: "item", item_id: 456, doc_name: "My Doc", column_id: "doc_col_1" , markdown: "..." } - Workspace doc with agent owner: { location: "workspace", workspace_id: 123, doc_name: "My Doc", markdown: "...", docOwnerIds: ["<agent_owner_user_id>"] }

create_folder

ChatGPT
Create a new folder in a monday.com workspace

create_form

ChatGPT
Create a monday.com form. Also creates a backing board to store responses. Returns the formToken for future mutations.

create_form_submission

ChatGPT
Submit a response to a monday.com WorkForm. Use get_form first to retrieve the WorkForm, then: - Inspect each question's showIfRules to determine which questions are conditionally shown based on previous answers. - Inspect each question's settings for any answer constraints (e.g. rating limits, select options, label limits). - Take note of any titles, descriptions, and content blocks to present the form naturally as you walk the user through it. - Take note of pages and question order to present questions in the correct sequence. Gather all answers upfront before calling this tool — do not submit one question at a time. Accepts a bare form token, a full WorkForm URL (e.g. https://forms.monday.com/forms/{form_token}?r=use1), or a shortened wkf.ms URL (e.g. https://wkf.ms/4tqP28t) — shortened URLs are automatically resolved by following the redirect. Returns the submission ID.

create_group

ChatGPT
Create a new group in a monday.com board. Groups are sections that organize related items. Use when users want to add structure, categorize items, or create workflow phases. Groups can be positioned relative to existing groups and assigned predefined colors. Items will always be created in the top group and so the top group should be the most relevant one for new item creation

create_item

ChatGPT
Create a single item or subitem on a monday.com board, or duplicate an existing item. [IMPORTANT] If you need to create multiple items in one call, use create_items instead of calling this tool in a loop. Otherwise: create a new item with provided values, create a subitem under a parent item, or duplicate an existing item and update it with new values. Use parentItemId when creating a subitem under an existing item. Use duplicateFromItemId when copying an existing item with modifications. [REQUIRED PRECONDITION]: Before using this tool, if new columns were added to the board or if you are not familiar with the board's structure (column IDs, column types, status labels, etc.), first use get_board_info with filters.columns.only to get column metadata without fetching views. This is essential for constructing proper column values and knowing which columns are available.

create_items

ChatGPT
Create up to 20 new items in a single call. Each item is fully independent - it chooses its own groupId, parentItemId (for subitems), duplicateFromItemId (for bulk templating from an existing item), and createLabelsIfMissing. A single call can therefore span multiple groups, mix regular items with subitems under different parents, and mix fresh creates with duplicates of existing items. Each item returns its own item_id and item_url on success, or a raw error message on failure. Batch each parent's subitems into groups of up to 20. Wait for one batch to finish before starting the next batch for that parent. [REQUIRED PRECONDITION]: Before using this tool, if new columns were added to the board or if you are not familiar with the board's structure (column IDs, column types, status labels, etc.), first use get_board_info with filters.columns.only to get column metadata without fetching views. This is essential for constructing proper column values and knowing which columns are available.

create_notification

ChatGPT
Send a notification to a user via the bell icon and optionally by email. Use target_type "Post" for updates/replies or "Project" for items/boards.

create_update

ChatGPT
Create a new update (comment/post) on a monday.com item. Updates can be used to add comments, notes, or discussions to items. You can optionally mention users, teams, or boards in the update. You can also reply to an existing update by using the parentId parameter.

create_updates

ChatGPT
Create up to 40 updates (comments/posts) in a single call. Each entry is independent - it targets its own item, has its own body and mentions, and can reply to an existing update via parentId - so one call can post on many items, or post several replies. Use this instead of calling create_update repeatedly when posting more than one update. Each entry returns its own update_id and item_id on success or a raw error message on failure.

create_view

ChatGPT
Create a new board view (tab) with optional filters and sorting. Filter operators: any_of, not_any_of, is_empty, is_not_empty, greater_than, lower_than, between, contains_text, not_contains_text View types: TABLE (standard board), DASHBOARD, FORM, APP

create_view_table

ChatGPT
Create a new table-type board view with optional filters, sort, tags, and table-specific settings including conditional coloring (highlight rows/cells based on column values). CONDITIONAL COLORING: Use settings.conditional_coloring to highlight rows or cells. Each rule specifies a column_id, operator, value (human-readable — e.g. "Stuck", not an index), color, and entire_row flag. Example: highlight rows where Status is "Stuck" in red, or where Salary > 100000 in green. Use this tool instead of create_view when you need table-specific settings like column visibility, group-by, or conditional coloring. Filter operators: any_of, not_any_of, is_empty, is_not_empty, greater_than, lower_than, between, contains_text, not_contains_text

create_widget

ChatGPT
Create a new widget in a dashboard or board view with specific configuration settings. This tool creates data visualization widgets that display information from monday.com boards: Parent Containers: - DASHBOARD: Place widget in a dashboard (most common use case) - BOARD_VIEW: Place widget in a specific board view Critical Requirements: 1. Schema Compliance: Widget settings MUST conform to the JSON schema for the specific widget type 2. Use all_widgets_schema first: Always fetch widget schemas before creating widgets 3. Validate settings: Ensure all required fields are provided and data types match Workflow: 1. Use 'all_widgets_schema' to get schema definitions 2. Prepare widget settings according to the schema 3. Use this tool to create the widget

create_workflow

ChatGPT
Creates a new empty workflow in the given workspace and returns its identifiers (workflowObjectId and workflowDraftId). Use this tool when the user wants to start a brand-new workflow from scratch, rather than modifying an existing one. Use it when the user asks for a workflow: a potentially robust process with extra capabilities like condition branching, multiple boards, or iterators. Only the workspace is required; title, privacy kind, description, folder, and owners are optional and fall back to sensible defaults. The tool returns a JSON object with the identifiers of the newly created workflow, which can then be used with the other workflow tools. If the get_workflow_editor_link tool is available, get the workflow's link from it instead of building one. Otherwise, to build a URL to the workflow, use the template: https://<account_slug>.monday.com/custom_objects/<workflowObjectId>. To get a real URL example of the account, call the monday GraphQL MCP tool with the query { me { url } }.

create_workspace

ChatGPT
Create a new workspace in monday.com

deactivate-sequence

ChatGPT
Deactivate a sequence so it stops accepting new enrollments. Deactivate a sequence (sets status to INACTIVE, blocking new enrollments while existing enrollments continue). Convenience alias for update-sequence-status when direction is known to be INACTIVE.

delete_action

ChatGPT
Use to permanently delete a saved action by its ID. Delete a saved action. Example: id: "550e8400-e29b-41d4-a716-446655440000"

delete_view

ChatGPT
Delete a board view (tab) from a monday.com board. Use get_board_info to find the view ID before deleting.

discover_public_brains

ChatGPT
List the public brains in the user's monday.com account that the user connected to this agent. Use it to find a brain to read without being given its id. Then read a brain with get_public_brain_files_index, passing its id as brainId, and its files with get_public_brain_file. Takes an optional cursor to fetch the next page.

duplicate-sequence

ChatGPT
Copy an existing sequence, including its steps, as a new INACTIVE sequence. Duplicate a sequence. Creates a copy with the same steps and configuration. The new sequence starts with INACTIVE status.

enroll-item-in-sequence

ChatGPT
Add CRM contacts or leads to an active outreach sequence. Enroll one or more CRM contacts or leads into an outreach sequence. The sequence must be active. Returns per-item success/failure so partial failures do not silently drop contacts.

execute_code

ChatGPT
Use to run arbitrary code (Python, JavaScript, TypeScript, or Bash) in a monday-authenticated sandbox for bulk operations, data transformation, file I/O, or multi-step workflows. Run arbitrary code in a monday-authenticated sandbox, without saving. Prefer dedicated monday tools for individual reads, writes, and GraphQL queries/mutations — they render in the UI and are retried one step at a time. Reach for execute_code when code is genuinely the better tool: - Bulk / multi-item work — batch operations, dedup, aggregations, joins across boards (one script beats N tool calls that accumulate context and compound failure) - Data transformation — normalizing phones/dates, fuzzy matching, weighted scoring - File I/O — parsing uploaded CSV/XLSX to import items, producing downloadable exports - Multi-step workflows where each step's output gates the next The sandbox has authenticated access to the monday.com API. You can make HTTP requests with GraphQL queries and mutations — authentication is handled automatically. IMPORTANT: Network access is restricted to the following hosts: [api.monday.com/, mcp.monday.com/mcp]. Requests to any other host (or a different path on a restricted host) will be blocked. Use this tool to query boards, items, columns, users, updates, and any other monday.com API resource. THE SANDBOX IS PER-CALL: a new empty container every call, destroyed when the call returns. Nothing written to disk survives, /tmp included. Never write a file in one call to read it in a later one, and don't invent staging paths for earlier tool results — none exist. To carry data forward, print it and pass it into the next call's code, or do the whole job in one call. Splitting a fan-out across calls only works if the later calls don't depend on the earlier ones' files. FAIL WITH A NON-ZERO EXIT. A run that prints an error and exits 0 is recorded as a success. The monday.com API returns HTTP 200 with an "errors" array, so check the parsed body rather than the status code and raise when it is present. Let exceptions propagate; don't wrap the script in a bare try/except. TIME LIMIT: 300s, or less if you pass timeout_sec. A run that exceeds its limit is killed, so scope each call to finish well inside the limit instead of fetching everything in one script. Don't call mcp.monday.com from inside the sandbox to reach monday tools — you already have them, and a tool missing from your tool list won't be found there either. Example — monday.com GraphQL query, raising on errors (Python). Use this shape for every API call: code: "import requests\ndef gql(query):\n body = requests.post('https://api.monday.com/v2', json={'query': query}).json()\n if 'errors' in body:\n raise RuntimeError(body['errors'])\n return body['data']\nprint(gql('{ users(limit:5) { id name email } }'))" Example — monday.com GraphQL mutation (Python): code: "import requests\nmutation = 'mutation { create_board(board_name: \"New Board\", board_kind: public) { id } }'\nbody = requests.post('https://api.monday.com/v2', json={'query': mutation}).json()\nif 'errors' in body:\n raise RuntimeError(body['errors'])\nprint(body['data'])" Example — with vars (accessed via os.environ): code: "import os, requests\nuser_id = os.environ['user_id']\nresp = requests.post('https://api.monday.com/v2', json={'query': f'{{ users(ids: [{user_id}]) {{ id name email }} }}'})\nprint(resp.json())" vars: {"user_id": 12345} Example — simple (Python): code: "print('hello world')" Example — with files (input and output): code: "import json\ndata = json.load(open('/tmp/data.json'))\njson.dump({'count': len(data)}, open('/tmp/result.json', 'w'))" files: [{"path": "/tmp/data.json", "content": "W3siaWQiOiAxfV0="}] output_files: ["/tmp/result.json"] Example — auto-collect outputs (write anything you want returned under /outputs): code: "import os, json\nos.makedirs('/outputs', exist_ok=True)\njson.dump({'ok': True}, open('/outputs/result.json', 'w'))" return_outputs: true

explore_meetings

ChatGPT
Find which meetings are about a topic, person or project, or list meetings by date; the starting point for questions across meetings. With a query, meetings are ranked by meaning across their AI-generated content (title, AI gist, summary, topics, action items; not the transcript) together with keyword overlap on the title and gist, so a paraphrase can still match. It always returns the closest meetings, even when none is really about the query: judge each by its title and gist. Results are ordered by relevance, not by date. A recurring meeting is several results with one title, one per occurrence. Without a query it lists meetings newest first, filtered by date and access. Read a meeting with get_meetings_content using its id.

extract_signals

ChatGPT
Answer what is going on inside a single CRM record — risks, objections, commitments and next steps drawn from its conversations. Extract structured CRM signals (summary, blockers, objections, buying signals, commitments, etc.) from communication resources for specific Monday item(s). Use when the user wants signals from a deal/contact's emails, meetings, or activities. Do not use for "top signals", account-wide priorities, or the CRM home signals widget — that is get_account_task_rollup (open account tasks). Single-item mode (itemId): returns stored signals or runs on-demand extraction. Batch mode (itemIds): returns only preprocessed stored signals per item; items without stored signals appear in itemsRequiringIndividualExtraction — re-call with a single itemId for on-demand extraction. Each resource returns { signals, message? }; unsupported types include an explicit "no signals yet" message. Each signal includes signalType, signalSubType, structuredData, and communicationEventsIds.

finalize_asset_upload

ChatGPT
Use to finish a file upload started with get_asset_upload_url and attach it to an item column. Finalize a file upload and create the asset on monday.com. Call this after uploading the file to the presigned URL from get_asset_upload_url. Requires the etag value from the PUT response headers. Automatically attaches the uploaded asset to the specified file column on the item. boardId is optional; when omitted, it is resolved from the item. Returns the created asset_id.

form_questions_editor

ChatGPT
Create, update, or delete a question in a monday.com form. [REQUIRED PRECONDITION]: For update and delete, call get_form first to resolve the exact question id and see its current type and settings — never guess a question id. For create, get_form shows the existing questions so you do not duplicate one.

get-activity-insights

ChatGPT
Fetch aggregated CRM activity stats: call counts and durations grouped by rep, activity type, or item. Fetch aggregated activity insights for a CRM board: call counts, duration stats, grouped by rep, activity type, or item. Does not return individual activity records.

get-board-sequences

ChatGPT
List the outreach sequences on a board or for the current user, with their status and enrollment counts. List sequences for a board or the current user. Returns active and inactive sequences (never deleted). Each entry includes status, enrollment counts, step count, and aggregate analytics. An empty list means no matching sequences.

get-contact-journey

ChatGPT
List every sequence a specific contact or lead is enrolled in, starting from that contact. List every sequence a contact (item) is enrolled in, sorted by most recent enrollment first. Returns each enrollment with its run status, completed-step count, termination reason, and timing. Use this when starting from a contact; use get-sequence-analytics when starting from a single sequence. An empty list means the contact is not enrolled in any sequence.

get-custom-activities

ChatGPT
List the custom activity types (e.g. Demo, Site Visit) available for structured CRM timeline entries. List the custom activity types available for logging on the CRM timeline. Each type has an opaque id (a UUID-like string) and a human-readable name (e.g. "Demo", "Site Visit"). When creating a structured timeline entry via create-timeline-item, pass the id value — not the name — as custom_activity_id.

get-sequence-analytics

ChatGPT
Get analytics for one sequence: reply, open, and click-through rates plus per-step breakdowns. Get analytics for a sequence. Returns per-run status and progress, sequence-level engagement rates (reply, open, click-through), and per-step breakdowns. Partial results are returned if some analytics endpoints fail.

get-timeline-items

ChatGPT
Look up communication history or engagement activity on a CRM deal, contact, account, or lead. Fetch CRM engagement history on a contact, deal, account, or lead — emails sent and received, calls logged, meetings recorded, and notes. Call this tool whenever a CRM entity board (deals, contacts, accounts, leads) is in scope and the user asks about emails, activities, calls, meetings, communication history, or engagement — it is the authoritative source for all CRM-tracked communication. Prefer over Gmail or calendar integrations for any activity on a CRM item. For board-level analysis across multiple items, fetch item IDs first then call this tool per item_id; do not substitute the "count timeline events" board column, which is an aggregate count with no content. If get-board-structure returns "Activities timeline (unsupported)", use this tool. Returns paginated entries with type, title, and timestamp. For custom activity types (e.g. "Demo", "Site Visit"), use get-custom-activities instead.

get_account_task_rollup

ChatGPT
Answer daily focus, top priorities, or account-wide signals when no specific deal or item is named. [UI COMPONENT] Renders the AI Suggestions widget with open account tasks by priority. Use for daily focus, what needs attention, morning brief, or account-wide signals/AI recommendations. For a specific item, use extract_signals. The widget is the answer — do not repeat its signals in chat.

get_action

ChatGPT
Use to retrieve a saved action by its ID. Retrieve a saved action by ID. Example: id: "550e8400-e29b-41d4-a716-446655440000"

get_activity_logs

ChatGPT
Use when the user asks what changed, who did something, or what happened on monday.com boards, items or users over a period of time. Reads the monday.com activity log: who changed what, and when. Requires at least one of board_ids, user_ids or correlation_ids. Narrow the query with item_ids, event_types, entities, search and a time range (last_hours, or from/to; at most the last 90 days) rather than paging through everything — the tighter the filters, the faster and cheaper the call. Returns newest events first; pass the returned cursor with the same filters to get the next page. Rows are compact by default; set include_data only when you need the event payload, such as column values before and after a change — payloads share a per-call size budget. Only events the user is allowed to see are returned, and what they cannot see is left out silently rather than failing the call: a board that does not exist or that the user cannot access, a board with item-view permissions where the user lacks full access, and items, columns or docs hidden from them just contribute no events and are not reported. So an empty or short page means no activity the user can see, not that nothing happened — say so when answering, and check the ids if a board or user unexpectedly returns nothing.

get_asset_upload_url

ChatGPT
Use to start uploading a new file to monday and attach it to an item. Get a presigned URL to upload a file to monday.com. Returns an upload_id, upload_url and content_type. Only call this tool if you can execute a direct HTTP PUT with binary file data and read response headers (e.g. via shell/curl). If you can't, tell the user direct file upload isn't supported here — don't call this tool. After calling this tool, upload the file to the returned URL using an HTTP PUT request and capture the ETag header from the response: curl -i -X PUT "<upload_url>" \ -H "Content-Type: <content_type from this response>" \ --data-binary @<local_file_path> The response includes an ETag header (e.g. ETag: "abc123...") — save this value. Then call finalize_asset_upload with uploadId (the upload_id), etag, itemId, columnId and optionally boardId to complete the upload and attach the file to an item's file column. Max file size: 500MB.

get_assets

ChatGPT
Use to get monday files (assets) by id, with their metadata and a download URL. Get assets (files) by their IDs. Returns file metadata including name, extension, size, public URL (valid for 1 hour), thumbnail URL, upload date, and who uploaded it. Takes asset ids, not item or board ids. An item's asset ids are the assetId values in its file column value.

get_assigned_items

ChatGPT
Returns items assigned to a user across their boards, optionally filtered by update date range. Defaults to the calling user.

get_automation_runs

ChatGPT
Read automation/workflow run history. Read-only. Modes: - "history": paginated run feed (state, duration, error reason). Use "filters" to narrow results and "nextPageOffset" to page (offset-only — next page = previous offset + returned count). - "detail": single run by "triggerUuid" (required) — returns block steps and MCP tool calls. Set "includeToolEvents": false to skip tool calls. Scope: provide "boardId" for a specific board or "accountWide": true. One is required. Known event states: "success", "failure", "exhausted".

get_automation_statistics

ChatGPT
Aggregate automation run statistics. Read-only. Breakdowns: - "totals": success/failure/total counts at the account or board level. - "by_entity": per-automation and per-workflow counts for a given "runStatus" (required: "success" | "failure" | "exhausted"). Use "excludeAutomationIds" to omit specific automations. Scope: provide "boardId" for a specific board or "accountWide": true. One is required. Optional "userIds" narrows results to specific creators.

get_board_activity

ChatGPT
Get board activity logs for a specified time range (defaults to last 30 days, up to the last 90 days). Optionally filter by item ids or user ids to avoid fetching activity for the entire board. [REQUIRED PRECONDITION]: Call this with includeData=true before undo_action — it is the source of the action_record_uuid that identifies the action to undo.

get_board_info

ChatGPT
Get comprehensive board information including metadata, structure, owners, and configuration. Set includeKnowledge to true when you need to interpret the board rather than just read its configuration: for example, choosing a status, label or column value from the user's intent instead of an exact name they gave, or explaining what statuses, columns or roles mean. Not needed for structural work such as column ids, types, revisions or views. Adds latency and may be null while knowledge is still being generated. Set includeSubscribers to true to also return the board's subscribers (people and AI agents) and team subscribers. Also returns the board's views (e.g. table views, filter views) — each view includes its id, name, type, and a structured filter object. On large boards, ALWAYS narrow the response: use filters.views.names or filters.views.ids when you only need specific views, and/or filters.columns.ids when you only need specific columns. Set filters.views.only or filters.columns.only when you want just that section — full views[].settings across many views can be multi-MB. The response includes hierarchy_type which indicates if the board is a multi-level board ("multi_level") where items can have nested subitems up to 5 levels deep on the same board. On multi-level boards, subitems share the same columns as parent items and subItemColumns will be null. Call this FIRST whenever you are not already familiar with a board structure (column IDs, column types, column revisions, status labels) — before reading or writing its data, or before any tool that declares this as a required precondition (e.g. get_board_items_page, board_insights, create_item, create_items, update_items, change_item_column_values, update_column, create_view, create_view_table, update_view, update_view_table). Also use the views it returns to resolve a view referenced by name (pass that name in filters.views.names), and as the source of view ids for update_view and update_view_table. Each column's "settings" field is the raw API value for that existing column, shown so you can read current labels/config — it is NOT the format expected by the columnSettings parameter of create_column or update_column. Never copy a column's "settings" object verbatim into columnSettings — use get_column_type_info with fetchMode "schema" to get the correct shape for the column type you are creating or updating.

get_board_items_page

ChatGPT
Get items from a monday.com board: list, filter, sort, search and paginate board items with their column values, groups, subitems and item descriptions. Returns structured JSON with board info (id, name, hierarchy_type), item details (id, name, url, created_at, updated_at, parent_item_id) and pagination (has_more, nextCursor, count). Use nextCursor from the response as cursor to get the next page while has_more is true. Filters, orderBy, itemIds and searchTerm are encoded in the cursor and ignored on cursor pages. [REQUIRED PRECONDITION]: Before using this tool, if new columns were added to the board or if you are not familiar with the board structure (column ids, column types, status labels, etc.), first use get_board_info with filters.columns.only to get column metadata without fetching views. Column ids are board-specific and are needed for filters, orderBy and columnIds. [FILTERS]: Each rule is {columnId, operator, compareValue} and optionally compareAttribute; filtersOperator ("and" by default) joins them. Before building a rule, call get_column_type_info with fetchMode "guidelines" for that column's type and follow data.guidelines.filter - the operator and the compareValue shape depend on the column type, and a wrong shape is rejected, not ignored. Invalid filters are rejected server-side with a structured error naming the offending rule and field, so check the guidelines first rather than guessing an operator. [VIRTUAL COLUMNS]: Four filterable and sortable columns exist that get_board_info never returns - "group" (the item's board group; the group id goes in compareValue), "__creation_log__" (creation time), "__last_updated__" (update time) and "__item_id__" (item id). All four are also valid in orderBy, e.g. "__creation_log__" with direction "desc" to sort newest-first. Their column types for get_column_type_info are not the ids: use columnType "group", "item_id", "creation_log" and "last_updated" respectively to get each one's compareValue and operator rules. [PAGE SIZE]: limit accepts up to 2000 items per page (default 25), so a large board can be read in far fewer calls than the previous 500 cap allowed. Ask for what you need: a bigger page costs one round trip but a larger response, and the few requests answered by the monday API (see the routing notes above) come back with 500 items, a cursor and a warning. [SEARCH]: Use searchTerm for vague or approximate wording (e.g. "marketing campaign"); use filters when the user gives an exact value for a specific column. A searchTerm result is capped at 20 items and cannot be paged: has_more true with nextCursor null means the result was capped, so do not repeat the call; use filters to read every match. [ITEM DESCRIPTION]: To retrieve an item description (the rich-text body/details of a monday.com item), set includeItemDescription to true - the response includes the description document blocks with their content, type and id. Use this whenever the user asks about an item description, body, details, or notes. [MULTI-LEVEL BOARDS]: The response includes hierarchy_type on the board ("multi_level" for MLS boards) and parent_item_id on each item. On multi-level boards items form a tree (up to 5 levels). Use includeSubItems to get descendants (each carrying parent_item_id to reconstruct the tree). Top-level items have no parent_item_id. Set includeIsLeaf to get is_leaf on each item (true when it has no subitems), e.g. to find the bottom-level tasks; it is already returned with includeColumns, includeSubItems or hierarchyScope. [REQUIRED PRECONDITION]: For board-relation / cross-board linking tasks, call link_board_items_workflow before using this tool. [VIEW-BASED FILTERING]: If the user refers to a board view by name (e.g. "show me items in the Overdue view"), first call get_board_info with filters.views.names set to that view name (avoids downloading all views on large boards), extract the matching view's filter field, then pass it as the filters argument here.

get_column_type_info

ChatGPT
Retrieves comprehensive information about a specific column type. Use fetchMode "schema" (default) to get the JSON schema definition from the API — to understand structure, validation rules, and available properties for column settings. Call this BEFORE any tool that writes column settings: create_column, update_column, and manage_object_schema_columns. Use fetchMode "guidelines" to get only guidelines.filter and guidelines.aggregation (no schema, no GraphQL round-trip). Call this before building any filter rule that uses compare_value/operator for that column type — e.g. get_board_items_page, board_insights, or a view's filters (create_view, create_view_table, update_view, update_view_table) — and before building board insights aggregation counts.

get_form

ChatGPT
Get a monday.com form by its form token, including its pages, questions, question ids, settings, and conditional showIfRules. Form tokens can be extracted from the form's url. Given a form url, such as https://forms.monday.com/forms/abc123def456ghi789?r=use1, the formToken is the alphanumeric string that appears right after /forms/ and before the ?. In the example, the formToken is abc123def456ghi789. Call this FIRST before any tool that acts on an existing form: create_form_submission (to know the questions and their constraints), form_questions_editor (to resolve question ids and current structure), and update_form (to see the current settings before changing them).

get_graphql_schema

ChatGPT
Fetch the monday.com GraphQL schema structure including query and mutation definitions. This tool returns available query fields, mutation fields, and a list of GraphQL types in the schema. You can filter results by operation type (read/write) to focus on either queries or mutations.

get_meeting_recording

ChatGPT
Get a temporary download link for one meeting recording. Bot and desktop recordings are video with audio; mobile recordings and phone calls are audio only. The link is private and expires: give it only to the user who asked, never post it to boards, updates, docs or other people, and call this tool again for a fresh link if it stops working. To read what was said, use get_meetings_content instead.

get_meetings_content

ChatGPT
Fetch full content (summary, topics, action items, transcript) for meetings you already have ids for. Get those ids from explore_meetings or search_meetings_content first; this tool is NOT for discovery or listing. Pass several ids in one call, with the include_ flags for the content you need (defaults to the summary if none are set). Requested ids that are not returned are listed at the end (not found, not accessible, or no completed recording); a meeting whose content was dropped to keep the response within its size limit says so, and can be requested again on its own. The search param is a narrow case-insensitive substring fallback on title, participant name, or email — NOT topic/keyword search. Quote numbers as the meeting states them, and don't attach a currency or unit the meeting didn't state.

get_monday_dev_sprints_boards

ChatGPT
Discover monday-dev sprints boards and their associated tasks boards in your account. Purpose: Identifies and returns monday-dev sprints board IDs and tasks board IDs that you need to use with other monday-dev tools. This tool scans your recently used boards (up to 100) to find valid monday-dev sprint management boards. What it Returns: - Pairs of sprints boards and their corresponding tasks boards - Board IDs, names, and workspace information for each pair - The bidirectional relationship between each sprints board and its tasks board Note: Searches recently used boards (up to 100). If none found, ask user to provide board IDs manually.

get_monday_knowledge

ChatGPT
Ask a question about monday.com and get an AI-generated answer from the official knowledge base. Use kind="general" for questions about using monday.com — features, automations, UI, help center, and settings. Returns cited source articles with links. Use kind="developer_docs" for questions about the monday.com API — GraphQL queries and mutations, authentication, rate limits, webhooks, schema, API best practices, and building apps. Important: do not include PII data in the questions.

get_public_brain_file

ChatGPT
Fetch one synthesized reference file (Markdown) from a public brain in the user's monday.com account. Pass the brainId of the brain and a fileKey from the "Generated Files" section returned by get_public_brain_files_index. Only public brains can be read.

get_public_brain_files_index

ChatGPT
Fetch the index (Markdown) of a public brain in the user's monday.com account: what the brain knows, its members, and its Generated Files with their file keys. Pass the brainId of the brain to read. Only public brains can be read. Then read a file with get_public_brain_file, passing the same brainId with a fileKey from Generated Files.

get_run_once_trigger_entities

ChatGPT
Lists the concrete entities (items, boards, docs, ...) the workflow's trigger can be fired on, so the workflow can be run once on one of them. Use this tool: - ALWAYS before calling run_workflow_once, to learn whether the trigger needs an entity and which entities are available. - When the user asks what the workflow would be run on. Do NOT use this tool for a board automation, such as one create_automation built: run-once applies to workflows only, and the id create_automation returns is not a workflowObjectId. Tell the user to trigger that automation on the board instead. The tool does NOT modify the workflow and does NOT run anything. The response is a JSON object: - status "ok": "requiresEntity" tells you whether run_workflow_once needs a triggerPayload, and "entities" is a list of { label, value } candidates. Show the labels to the user and let them pick; pass the picked entity's "value" object verbatim as run_workflow_once's triggerPayload. An empty "entities" list while requiresEntity is true means there is nothing to run on — tell the user instead of guessing a payload. - status "failed": "message" explains why the entities could not be listed. Relay it; do not call run_workflow_once.

get_sprint_summary

ChatGPT
Get the complete summary and analysis of a sprint. Purpose: Unlock deep insights into completed sprint performance. The sprint summary content including: - Scope Management: Analysis of planned vs. unplanned tasks, scope creep - Velocity & Performance: Individual velocity, task completion rates, workload distribution per team member - Task Distribution: Breakdown of completed tasks by type (Feature, Bug, Tech Debt, Infrastructure, etc.) - AI Recommendations: Action items, process improvements, retrospective focus areas Requirements: - Sprint must be completed and must be created after 1/1/2025 - Requires a sprintId. [REQUIRED PRECONDITION]: call get_sprints_metadata first to list the sprints on the board and resolve the sprintId (and to confirm the sprint is completed). If you do not know the sprints board ID either, start with get_monday_dev_sprints_boards. Important Note: When viewing the section "Completed by Assignee", you'll see user IDs in the format "@user-12345678". the 8 digits after the @is the user ID. To retrieve the actual owner names, use the list_users_and_teams tool with the user ID and set includeTeams=false for optimal performance.

get_sprints_metadata

ChatGPT
List the sprints of a monday-dev sprints board with their metadata. Returns comprehensive sprint metadata including: Data Retrieved: A table of sprints with the following information: - Sprint ID - Sprint Name - Sprint timeline (planned from/to dates) - Sprint completion status (completed/in-progress/planned) - Sprint start date (actual) - Sprint end date (actual) - Sprint activation status - Sprint summary document object ID Parameters: - limit: Number of sprints to retrieve (default: 25, max: 100) Requires the Main Sprints board ID of the monday-dev containing your sprints. If you do not already have it, call get_monday_dev_sprints_boards first to discover it. Call this before: - get_sprint_summary — this tool returns the Sprint IDs that get_sprint_summary requires. Never guess a sprint ID.

get_type_details

ChatGPT
Get detailed information about a specific GraphQL type from the monday.com API schema, including its fields, input fields, and enum values. Call this with a known type name to confirm its exact fields, arguments, and enum values before referencing that type in an operation for all_monday_api, all_api_read, or all_api_write, so the fields and arguments you send actually exist.

get_updates

ChatGPT
Get updates (comments/posts) from monday.com items or a board. Pass the IDs in objectIds and set objectType: up to 25 item IDs for Item, in one call instead of calling once per item, or one board ID for Board. Filter by date range with fromDate and toDate (both required together, ISO8601 format). By default, Board queries return only board discussion. Set includeItemUpdates to true to also include updates on individual items, and add a date range to keep the response small. Returns update text (bodies over 2000 characters are truncated and flagged with text_body_truncated. To read one in full, call again with only that item's ID and includeFullText), creator info, timestamps, and optionally replies and assets.

get_user_context

ChatGPT
Fetch current user information, account information, and their relevant items (boards, docs, folders, workspaces, dashboards). Use this tool to: - Get context about who the current user is (id, name, title) - Get account info: plan tier, active member count, trial status, and active products - Get the number of active members in the account (returns active_members_count) - Discover user's favorite boards, folders, workspaces, and dashboards - Get user's most relevant boards based on visit frequency and recency - Get user's most relevant docs based on load frequency and recency - Get user's most relevant people based on interaction frequency and recency - Reduce the need for search requests by knowing user's commonly accessed items

get_user_mentions

ChatGPT
Returns recent updates and comments the user was @mentioned on, with the post content, most recent first. Defaults to the calling user.

get_user_recent_activity

ChatGPT
Returns a chronological timeline of what one or more users did across their boards (created, updated, completed, deleted, renamed, moved items). To list distinct tasks worked on, group the timeline by item.id; to list active boards, group by board.id. Defaults to the calling user. Note: when "completed" or columnType filters are used, results are drawn from the user's most recent activity within the time window, so increase "from"/"limit" if you expect older matches.

get_workflow_run_once_status

ChatGPT
Reports how a run started by run_workflow_once ended: whether it succeeded, and if not, which step failed and why. Call this after run_workflow_once, passing the "automationId" it returned. Pass the same workflowDraftId you passed to run_workflow_once. The tool does NOT modify the workflow and does NOT run anything. It answers immediately with whatever is known so far, so a run in progress is a normal answer, not a problem: - "isTerminal": false — the run has NOT finished. This is NOT a failure. Wait "retryAfterMs" milliseconds and call this tool again with the same arguments. Keep doing that until isTerminal is true. Never report an outcome to the user while isTerminal is false; say the run is still going. - "isTerminal": true — the run is over and "state" is its final outcome. Report it and stop calling. Trust "isTerminal" over your own reading of "state". "state" is one of: - "success" — every step ran without error. - "failure" — a step failed. "errorReason" and the failing entry in "blocks" say which and why. - "exhausted" — the run gave up after retrying a step too many times. - "stopped" — the run was stopped before finishing, so the remaining steps never ran. - "zero_actions" — the trigger fired but the workflow's conditions matched nothing, so no step ran. Nothing is broken; the entity did not qualify. - "running" — a step is executing right now. - "waiting" — the run is parked on a wait step or waiting for a second event. This can last hours, which is why "retryAfterMs" is long here. Tell the user it is waiting rather than polling silently for hours. - "not_indexed_yet" — the run's record has not appeared yet, which is expected for the first few seconds after a run starts. It does NOT mean nothing ran. "blocks" lists the steps recorded so far, each with its own "state" and, when it failed, "errorReason". While isTerminal is false this list is partial and is progress, not a verdict. Use it to tell the user which step failed rather than only that the run failed. The response is a JSON object: - status "ok": the fields above. - status "failed": the status could not be read. "message" explains why. This says nothing about the run itself — do not report the run as failed because of it.

invoke_process_planner

ChatGPT
A reasoning-focused process planner with deep knowledge of monday.com workflow architecture. Given a description of a process, it returns a structured textual plan describing one or more related workflows that implement it. Use this tool for: - Planning a new workflow or multi-workflow architecture from a process description. - Deciding whether a process should be implemented as a single workflow or multiple related workflows. - Producing a concrete, block-level plan (trigger, steps, route keys) that another agent or human can build from. - Don't tell it about one time actions, like non-repeating resource creations that are needed to set up the process. This tool does NOT have access to any specific workflow's current state — it plans from scratch using all available blocks. It does not execute any changes.

invoke_workflow_expert

ChatGPT
Workflow expert for a single workflow. Given a prompt, answers questions about the workflow's structure and configuration, or makes changes to it (create, update, delete steps, and configure step fields). Delegate any prompt that asks about a workflow or asks to change it. Pass clear, descriptive instructions — the expert will decide the right response or operations. Field values reference resources (boards, columns, people, channels, projects, ...) from monday or any external app — all handled the same way. Pass each resource through as the user stated it; the expert resolves names to IDs and asks the user when it's ambiguous. Pro-tip: pass along any ID you already have to save a follow-up question — no need to look one up first. Constraints: works on ONE workflow at a time.

list_action_items

ChatGPT
List or search action items (tasks, follow-ups, commitments) across meetings. Without a query it lists every action item matching the filters, newest meeting first, e.g. "all my action items from last week" or "what does Dana owe from the retrofit syncs". With a query it matches by meaning as well as words. Each item shows its text, owner and due date as the notetaker recorded them, under its meeting's title, date and id. Owners are names the notetaker wrote down, not monday users, so an item nobody explicitly took on has no owner. Done/open status is not shown: to check it, open the meeting with get_meetings_content, which also reads a full meeting.

list_actions

ChatGPT
Use to list all saved actions for the current user. List all saved actions for the current user.

list_agents_in_account

ChatGPT
Read the agents of the whole account, including agents the caller is not a member of. READ-ONLY. For account admins only: a caller without permission to manage account agents gets a PERMISSION_DENIED error. Use this for admin questions such as "which agents does my account have", "who owns or can use agent X", "which agents are inactive", "which agents have no subscribers" or "which agents are high risk". For the caller's own agents, use the normal agent tools instead. MODES: - List: no ids. Returns one page of the account's agents, newest first, with next_cursor while more remain. Pass next_cursor as cursor for the next page. - Lookup: ids and/or agent_user_ids. Returns those agents; unknown or deleted ones are left out. Each agent has id, name, role, kind, state, agent_user_id, owner_user_id, goal (first 200 characters) and timestamps. Set include_subscribers to also get who can use it: users and teams, each with a role of OWNER, EDITOR or MEMBER. A team id of -1 means everyone in the account. Set include_resources to also get the boards and docs connected to each agent, as resource_id and type (board or document), without access level. Set include_risk to also get each agent's governance risk: level High, Medium or Low, from the sensitivity of the data it reaches and the reach of its tools, and a score from 0 to 1. Set include_credits_used to also get the AI credits each agent consumed, in the current billing cycle by default or in the dates you pass.

list_automations

ChatGPT
List all automations on a specific monday.com board, including their ids, titles, active state, and configuration. Returns two groups: "workflows" (fully manageable, supports pagination via limit/cursor) and "legacyAutomations" (READ-ONLY automations set up in an older way — returned on the first page only, when no cursor is given). Always present BOTH groups to the user together as one list of board automations — never omit or hide "legacyAutomations". The split and the read-only limitation are internal context for you only: don't volunteer them and don't label any automation "legacy" or "read-only". Only if the user asks to act on a legacy one, briefly explain it can't be changed (e.g. "This one was set up in an older way and can't be modified here"). Call this FIRST before manage_automations — it is the only way to resolve an automation id from a name or description. Never guess an automation id. When NOT to use: Do not call this tool to get general board information unrelated to automations.

list_users_and_teams

ChatGPT
Tool to fetch users and/or teams data. MANDATORY BEST PRACTICES: 1. ALWAYS use specific IDs or names when available 2. If no ids available, use name search if possible (USERS ONLY) 3. Use 'getMe: true' to get current user information 4. AVOID broad queries (no parameters) - use only as last resort REQUIRED PARAMETER PRIORITY (use in this order): 1. getMe - STANDALONE 2. userIds 3. name - STANDALONE (USERS ONLY, NOT for teams) 4. teamIds + teamsOnly 5. No parameters - LAST RESORT CRITICAL USAGE RULES: • userIds + teamIds requires explicit includeTeams: true flag • includeTeams: true fetches both users and teams, do not use this to fetch a specific user's teams rather fetch that user by id and you will get their team memberships. • name parameter is for USER search ONLY - it cannot be used to search for teams. Use teamIds to fetch specific teams.

list_workspaces

ChatGPT
List all workspaces available to the user, ordered by membership (user's workspaces first). Returns workspaces with their ID, name, and description. [IMPORTANT] To search for workspaces by name, use the "search" tool with searchType WORKSPACES instead — it provides faster and more accurate results.

manage_agent

ChatGPT
Full lifecycle management for monday platform agents — create, read, update, delete, change state, and run. monday platform agents are user-built work orchestrators on monday.com. Each has a profile, goal, and agent-level Identity. Jobs define specific work, instructions, and triggers. Agents in state ACTIVE can be triggered automatically. They are NOT local LangChain or MCP agents. ACTIONS (only pass fields that apply to the chosen action): - create: { action:"create", prompt, identity?, agent_model?, skill_ids? } — AI-generated agent. Platform creates profile, goal, and Identity from the prompt unless identity is supplied. - create_blank: { action:"create_blank", name?, role?, role_description?, avatar_url?, gender?, background_color?, user_prompt? } — manually defined agent. - get one: { action:"get", agent_id } - list owned: { action:"get" } - update: { action:"update", agent_id, name?, role?, role_description?, identity?, plan?, agent_model? } - delete: { action:"delete", agent_id } - activate: { action:"activate", agent_id } - deactivate: { action:"deactivate", agent_id } - run: { action:"run", agent_id } RULES: - "create_blank" with no fields creates a nameless blank agent — only do this intentionally. - "update" requires at least one of name/role/role_description/identity/plan/agent_model. - Do not put job-specific instructions in Identity. Configure them with manage_agent_jobs. - "update", "delete", "activate", "deactivate", "run" all require "agent_id". - Created agents start INACTIVE. Follow with action:"activate" using the returned agent_id before they can be triggered. - ⚠️ DESTRUCTIVE — "delete" is permanent and irreversible. When the user refers to an agent by name, ALWAYS call action:"get" first to confirm the correct agent_id before deleting. - "run" is fire-and-forget. Returns trigger_uuid — no run-status query exists, treat successful enqueue as the only signal. - Agent state is one of ACTIVE, INACTIVE, ARCHIVED, or FAILED. DELETED only appears as the return value of action:"delete". USAGE EXAMPLES: - AI create: { "action": "create", "prompt": "Run my daily standup every weekday at 9am." } - Manual create:{ "action": "create_blank", "name": "Standup Bot", "role": "Project Manager", "gender": "female" } - Fetch one: { "action": "get", "agent_id": "42" } - List mine: { "action": "get" } - Rename: { "action": "update", "agent_id": "7", "name": "New Name" } - Activate: { "action": "activate", "agent_id": "7" } - Deactivate: { "action": "deactivate", "agent_id": "7" } - Run: { "action": "run", "agent_id": "7" } - Delete: { "action": "delete", "agent_id": "7" } RELATED TOOLS: - agent_catalog — browse available trigger types and skills before wiring them to an agent - manage_agent_jobs — configure jobs, job instructions, and nested triggers - manage_agent_triggers — manage which triggers fire this agent automatically - manage_agent_skills — manage which skills this agent can perform - manage_agent_knowledge — manage which boards/docs this agent has access to

manage_agent_jobs

ChatGPT
Manage jobs on a monday platform agent. Jobs define WHAT work runs and own one or more triggers that define WHEN it runs. ACTIONS: - list: { action:"list", agent_id } — returns jobs, their instructions, stable ids, and nested triggers. - create: { action:"create", agent_id, name?, instructions?, enabled?, triggers?, model?, thinking_level? } - update: { action:"update", agent_id, job_id, name?, instructions?, enabled?, triggers?, remove_trigger_ids?, model?, thinking_level? } - delete: { action:"delete", agent_id, job_id } — removes the job and all nested triggers. - run: { action:"run", agent_id, job_id } — runs the job once now, regardless of its triggers. RULES: - Agent-level Identity, tone, and guardrails belong in manage_agent. Job-specific work belongs here. - Call agent_catalog action:"list_triggers" before adding a trigger. Never guess block_reference_id. - trigger_id is the stable CRUD identity. node_id is only the workflow position. - Updates are patches: omitted jobs and triggers remain unchanged. - Credentialed third-party triggers cannot be configured here. Complete those trigger connections in monday.com. - action:"run" is fire-and-forget: a returned trigger_uuid means the run was accepted, not finished. - Job creation and trigger registration are not atomic with agent creation. If this tool fails after manage_agent created an agent, keep the agent and retry this tool. RELATED TOOLS: - manage_agent — create the agent and configure its agent-level Identity - agent_catalog — discover supported trigger types and required fields - manage_agent_triggers — legacy flat-trigger compatibility

manage_agent_knowledge

ChatGPT
List, grant, update, or revoke a monday platform agent's access to boards and docs. An agent's "knowledge" is the set of monday.com boards and docs it can read from or write to during a run. - list: Returns all resources the agent currently has access to, including permission level and resource type. - add: Grants the agent access to a board or doc with the specified permission level. - update: Changes the permission level on a resource the agent already has access to. Call action:"list" first to confirm the resource_id exists. - remove: Revokes the agent's access to a board or doc entirely. Call action:"list" first to confirm the resource_id exists. Permission types: - READ: Agent can read data from the resource. - READ_WRITE: Agent can read and write data to the resource. USAGE EXAMPLES: - List: { "action": "list", "agent_id": "7" } - Add board access: { "action": "add", "agent_id": "7", "resource_id": "42", "scope_type": "BOARD", "permission_type": "READ" } - Update to read-write: { "action": "update", "agent_id": "7", "resource_id": "42", "scope_type": "BOARD", "permission_type": "READ_WRITE" } - Remove access: { "action": "remove", "agent_id": "7", "resource_id": "42", "scope_type": "BOARD" } RELATED TOOLS: - manage_agent — manage the agent entity itself (create, activate, deactivate, etc.) - manage_agent_triggers — manage which triggers fire this agent automatically - manage_agent_skills — manage which skills this agent can perform

manage_agent_skills

ChatGPT
Manage the full skill lifecycle for monday platform agents — create new skills in the catalog, attach skills to an agent, or detach them. Skills extend what an agent can do (e.g. sending emails, querying databases, posting to Slack). ACTIONS: - create: { name, content, description? } — creates a new custom skill in the account-wide catalog. The skill becomes available to all agents in the account. - add: { agent_id, skill_id } — attaches a skill to this agent. - remove: { agent_id, skill_id } — detaches a skill from this agent. WORKFLOW — attach an existing skill: 1. Call agent_catalog action:"list_skills" — find the skill_id of the skill to attach. 2. Call this tool action:"add" with agent_id and that skill_id. WORKFLOW — create a new skill and attach it: 1. Call this tool action:"create" with name and content — note the returned id. 2. Call this tool action:"add" with agent_id and that id directly (no catalog lookup needed). NOTE: There is no action to list which skills are currently attached to a specific agent — the platform does not yet expose that query. To browse all skills available in the account catalog, use agent_catalog action:"list_skills". USAGE EXAMPLES: - Create a skill: { "action": "create", "name": "Send Slack Message", "content": "## Instructions\nPost a message to a Slack channel.", "description": "Sends a message to Slack" } - Add a skill: { "action": "add", "agent_id": "7", "skill_id": "skill-abc-123" } - Remove a skill: { "action": "remove", "agent_id": "7", "skill_id": "skill-abc-123" } RELATED TOOLS: - agent_catalog action:"list_skills" — browse existing skills to find a skill_id before calling action:"add" - manage_agent_triggers — manage which triggers fire this agent automatically - manage_agent — manage the agent entity itself (create, activate, deactivate, etc.)

manage_agent_tools

ChatGPT
Add integration tools (Gmail, Slack, ...) and built-in capabilities (web search, notetaker, ...) to a monday platform agent. ACTIONS: - search: { app_name?, query? } — lists integration tools with their block_id and whether they need an account, and always lists built_in_capabilities with their keys. With neither app_name nor query it lists the available app names instead. - add: { agent_id, block_ids?, built_in? } — attaches those tools and capabilities to the agent's draft. At least one of block_ids or built_in. WORKFLOW: 1. Call action:"search" with an app name (or a keyword such as "sms" when the app is unknown) and pick the block_ids (or built-in keys) that match what the agent needs. 2. Call action:"add" with agent_id and those ids. Add tools of one app per call. ADD RESULT: - added: the tool or capability is on the agent and switched on. - needs_connection: saved but switched off, no account linked. The user connects an account and switches it on in the agent's Tools panel in monday agents. - needs_activation: saved and linked to the user's account, but switched off. The user reviews its settings and switches it on in the agent's Tools panel in monday agents. - skipped: not added, with a reason code (already_added, already_enabled, pending_activation, not_found, turned_off, not_available, needs_user_confirmation) and a next_step when the user has to act. Only tools under added are ready. Do not tell the user any other tool is ready. LIMITS: - Tools that need an account (needs_account: true in search) cannot be added to a shared agent through this tool. - image_generation spends account credits, so it is never switched on here: it comes back under skipped, and the user turns it on in the agent's settings. - Removing tools is not supported here. USAGE EXAMPLES: - Search by app: { "action": "search", "app_name": "gmail" } - Search by keyword: { "action": "search", "query": "sms" } - Add: { "action": "add", "agent_id": "7", "block_ids": [123456] } - Add a built-in: { "action": "add", "agent_id": "7", "built_in": ["web_search"] } RELATED TOOLS: - manage_agent — manage the agent entity itself - manage_agent_skills — attach skills (instructions) rather than integration tools

manage_agent_triggers

ChatGPT
Legacy flat-trigger management for a monday platform agent. When jobs with instructions are enabled, use manage_agent_jobs so each trigger belongs to an explicit job. ACTIONS: - list: { agent_id } — returns active triggers with node_id, block_reference_id, name, field_summary. - add: { agent_id, block_reference_id, field_values? } — attaches a trigger type to the agent. - remove: { agent_id, node_id } — detaches a trigger instance by node_id (NOT block_reference_id). WORKFLOW — add a trigger: 1. Call agent_catalog action:"list_triggers" — note block_reference_id, field_schemas, and required_fields. 2. Collect required field values from the user (e.g. board_id, column_id). 3. Call this tool action:"add" with block_reference_id and field_values. Note: add returns only { success } — no node_id for the new instance. Call action:"list" afterward if you need the node_id. WORKFLOW — remove a trigger: 1. Call action:"list" to see active triggers and note the node_id of the instance to remove. 2. Call action:"remove" with that node_id. NOTE: Only triggers that can be added programmatically appear in the catalog. OAuth/3rd-party triggers (Slack, Gmail, Salesforce, etc.) require user setup in the monday.com UI — they will not appear in agent_catalog and cannot be managed here. USAGE EXAMPLES: - List triggers: { "action": "list", "agent_id": "7" } - Add trigger: { "action": "add", "agent_id": "7", "block_reference_id": "status-change-ref", "field_values": { "board_id": "42" } } - Remove trigger: { "action": "remove", "agent_id": "7", "node_id": "node-abc" } RELATED TOOLS: - manage_agent_jobs — jobs-aware configuration with per-job instructions and nested triggers - agent_catalog action:"list_triggers" — discover available trigger types and their required field_values before calling action:"add" here - manage_agent_skills — manage which skills this agent can perform - manage_agent — manage the agent entity itself (create, activate, deactivate, etc.)

manage_automations

ChatGPT
Activate, deactivate, or delete an existing monday.com automation. Requires an automation id. When the user refers to an automation by name, always call list_automations first to resolve the id — never guess or infer ids. Actions: - activate: enables a paused automation so it starts responding to its trigger. - deactivate: pauses an automation while preserving its definition. - delete: permanently removes an automation — irreversible. When intent is ambiguous ("stop", "turn off", "pause"), prefer deactivate over delete.

move_object

ChatGPT
Move a folder, board, or overview in monday.com. Use position for relative placement based on another object, parentFolderId for folder changes, workspaceId for workspace moves, and accountProductId for account product changes.

propose-crm-setup

ChatGPT
Set up, configure, or design a CRM of any kind (sales pipeline, clients, leads, deals, contacts, accounts) in monday.com, including a bare "set up a CRM" request or a question about which CRM setup fits a business. Propose a CRM board layout for a monday.com account — which boards to create, their purpose, pipeline stages, and tracked field areas — in the user's own vocabulary. Covers setting up, configuring, or restructuring a CRM of any kind: sales pipelines, client, lead, or deal tracking, industry CRMs such as real estate, recruiting, agency, or fundraising, and the structure of a CRM used elsewhere such as HubSpot or Salesforce. Also answers which CRM setup fits a business. Read-only: does not create or modify boards, items, or columns; build-crm-setup does that after the user has seen the proposal and approved it in a later message. Works from no context at all: a bare "set up a CRM" is enough to start. The tool runs its own short interview and designs the whole board set, so no clarifying questions beforehand and no board-by-board assembly with individual board tools are needed. Does not compare CRM vendors: "what is the best CRM?" is a product question, not a setup request, and a tailored setup can follow once the user wants one. Does not find or source new leads. Returns NEEDS_INFO with exactly one targeted question when the context is too thin for a useful proposal, the expected first result for a bare request; caps at three questions per conversation. Returns PROPOSAL with a board list and a setup_token once there is enough context, plus warnings for defaults used in place of missing answers and for common setup mistakes. To change a proposal the user has seen, call it again with its setup_token and refinements. A setup_token that expired within the last week still carries the answers, and the plan is regenerated for approval. Returns INVALID_TOKEN, without a token, when the supplied setup_token cannot be used; omit it to restart. Returns CRM_REQUIRED when monday CRM is not enabled on the account. Scope: this tool builds CRM structure only. It does not import records from another CRM, build pure project-delivery workflows, model nested client-of-client accounts, split one motion into product or regional pipelines, enforce per-assignee privacy, or undo created boards. Saved views, automations, and mirror columns are not created by this version. Phone-first and field-sales teams can use the boards but should skip the optional email connection. Email and calendar consent must be completed by the user in a browser. Importing contacts from Gmail into the Contacts board is optional, offered after the boards are built, and happens only if the user agrees to it. Use individual board tools or a manual setup for pure project delivery and nested client-account structures.

read_docs

ChatGPT
Get information about monday.com documents. Supports two modes: MODE: "content" (default) — Fetch documents with their full markdown content. - Requires: type ("ids" | "object_ids" | "workspace_ids") and ids array - For a doc that lives as a board view, pass board_id and view_id instead of type/ids. They are the ids in the board URL: https://<slug>.monday.com/boards/<board_id>/views/<view_id>. Omit view_id to read every doc view of the board. Such docs cannot be found by type "ids"/"object_ids". - Supports pagination via page/limit. Check has_more_pages in response. - If type "ids" returns no results, automatically retries with object_ids. - Set include_blocks: true to include block IDs, types, and positions in the response — required before calling update_doc. - Blocks default to 25 per page. Use blocks_limit and blocks_page to paginate through long documents. - Set include_comments: true to fetch all comments and replies on the document. Each comment is enriched with anchor info (block_id, selection_from, selection_length) indicating which block and text range it's attached to. Use comments_limit to control how many comments per item (default 50). MODE: "version_history" — Fetch the edit history of a single document. - Requires: ids with the document's object_id (use the object_id field from content mode results, NOT the id field). - The object_id is the numeric ID visible in the document URL. - Returns restoring points sorted newest-first. Use version_history_limit to cap results (e.g., "last 3 changes" → version_history_limit: 3). - Use since/until to filter by time range. If omitted, returns full history. - Set include_diff: true to see what content changed between versions (fetches up to 10 diffs, may be slower). - Examples: - { mode: "version_history", ids: ["5001466606"], version_history_limit: 3 } - { mode: "version_history", ids: ["5001466606"], since: "2026-03-11T00:00:00Z", include_diff: true }

run_action

ChatGPT
Use to execute a previously saved action by its ID, optionally passing variables. Execute a saved action by ID. Optionally pass variables (injected as environment variables, access via os.environ). Example: id: "abc-123", vars: {"board_id": 12345, "limit": 5}

run_workflow_once

ChatGPT
Runs the workflow once: a single execution, right now, on one real entity, without publishing or activating the workflow. When workflowDraftId is given the draft revision runs; otherwise the published live revision runs. THIS PERFORMS REAL SIDE EFFECTS. The blocks act on real boards, items and third-party apps: items get created and updated, notifications and emails go out, external systems are called. Nothing is simulated and nothing is rolled back. Required sequence — do not skip a step: 1. Call get_run_once_trigger_entities and show the user the entity labels. 2. Ask the user to confirm the run explicitly, naming the entity it will run on and warning that the effects are real. Never run on your own initiative, and never pick the entity for the user. 3. Only after the user confirms, call this tool with that entity's "value" as triggerPayload (omit triggerPayload when get_run_once_trigger_entities reported requiresEntity: false). The response is a JSON object: - status "running": the run was accepted and is now executing. You do not know its outcome yet, so never claim it succeeded, failed, or what it produced. Call get_workflow_run_once_status with the "automationId" from this response to find out; the user can also watch per-step progress in the builder UI. - status "validation_failed": the draft is not fully configured, so nothing ran. "issues" lists what to fix; help the user fix them, then start over from step 1. - status "failed": the run could not be started. "message" explains why — commonly this workflow's trigger cannot be run once. Relay the message; do not retry blindly.

search-crm-recipients

ChatGPT
Find CRM contacts, leads and people with an email address to send an email or mass email to. Find the people to email on the user's boards that have an email column (contacts, leads, accounts and others), by name or email address. Returns the boards with an email column and the matching people that have an address, grouped by board. Does not draft or send anything.

search_meetings_content

ChatGPT
Find where specific words appear inside meeting content and return the matching passages, across many meetings at once. Matches keywords only, not meaning, in discussed topics, the AI summary and action items: names, terms, codenames and numbers work best. Each result shows the meeting's title, date and AI overview, then its matching passages. The overview is a short gist for telling meetings apart: it leaves out most figures, owners, quotes and corrections, which are in the passages and in get_meetings_content. Transcripts are not searched. Without a query it returns content filtered by date/access. Read a meeting with get_meetings_content using its id. Quote numbers as the meeting states them, and don't attach a currency or unit the meeting didn't state.

send-mass-emails

ChatGPT
Send the mass email the user approved in the email draft widget. Send the email the user approved in the email draft widget, one copy per recipient, from the user's connected mailbox. Only the widget calls this, when the user clicks Send. Never call it yourself; to email people, use show-mass-emails-draft.

show-ai-suggestion-event

ChatGPT
Show a single AI suggestion (CRM task) as a card, e.g. after a new-suggestion event. [UI COMPONENT] Shows one CRM AI suggestion as a card: its action, the CRM item with its owners and linked account, contacts or deals, a link to My Focus and, for follow-ups, a button to draft the follow-up. Use when a crm.ai_suggestion_created event arrives (pass its task_id) or the user asks to see a specific suggestion. The card is the answer — do not repeat it in chat.

show-assign

ChatGPT
Use for requests to see or use an interactive assignment interface. [UI COMPONENT] Renders an interactive smart assignment interface visualization that the user can see and interact with. IMPORTANT: This is a UI DISPLAY tool - use it to RENDER visual components for the user to see and interact with. Do NOT use data-fetching tools when the user explicitly asks to "show", "display", "visualize", or "see" something visually. Helps assign tasks to the right people. Assignment suggestions are based on task details (like name) and person details (such as title, availability, etc). Always show as much as data possible, while showing the person details like title etc. If you do not have the data available - use the list_users_and_teams tool.

show-battery

ChatGPT
Use when user asks for: battery view, progress indicator, status distribution bar, completion percentage visualization, or Monday.com style status breakdown. [UI COMPONENT] Renders an interactive battery/progress indicator visualization that the user can see and interact with. IMPORTANT: This is a UI DISPLAY tool - use it to RENDER visual components for the user to see and interact with. Do NOT use data-fetching tools when the user explicitly asks to "show", "display", "visualize", or "see" something visually.

show-chart

ChatGPT
Use when user asks for: pie chart, bar chart, line graph, data visualization, or any graphical representation of numbers/statistics. [UI COMPONENT] Renders an interactive chart/graph visualization that the user can see and interact with. IMPORTANT: This is a UI DISPLAY tool - use it to RENDER visual components for the user to see and interact with. Do NOT use data-fetching tools when the user explicitly asks to "show", "display", "visualize", or "see" something visually.

show-mass-emails-draft

ChatGPT
Draft and send an email, or a mass email, to CRM contacts, leads and people. Show an editable draft of an email or a mass email to one or more CRM contacts, leads or other people the user manages, including bulk emails, campaigns and email blasts. Emails go out from the user's own Gmail or Outlook mailbox when they click Send; this is the only way to send email to CRM contacts. Pass the monday item ids of anyone the user named (found via search-crm-recipients), and a subject and body when you can write them from the conversation; the user can add or remove recipients from their CRM in the widget. Handles mailbox connection itself: when no Gmail or Outlook mailbox is connected, the widget asks the user to connect one first. The user reviews and edits the recipients, subject and body in the widget, then sends or declines it there. This tool does not send anything and cannot tell you whether the user sent the emails.

show-monday

ChatGPT
Open the full monday.com web app in the ChatGPT sidebar. Opens the full monday.com web app for the current account. Launched from the ChatGPT sidebar.

show-table

ChatGPT
Use when user asks to: display a board as table, show items in table format, view data in tabular layout, or see a Monday.com board visually. [UI COMPONENT] Renders an interactive table visualization that the user can see and interact with. IMPORTANT: This is a UI DISPLAY tool - use it to RENDER visual components for the user to see and interact with. Do NOT use data-fetching tools when the user explicitly asks to "show", "display", "visualize", or "see" something visually. When asked to update an item, use the currently selected item ID (get it from the widget state, using tools like "get_widget_state") for deciding which item to update. If no item is selected, ask the user which item should be updated. After adding an update to an item, you MUST display the table AGAIN, even if the user did not ask you to. [IMPORTANT][FILTERING PRECONDITION]: IF using filters, you MUST call get_board_info(boardId) FIRST and use the returned boardContextToken.

show-table-preview

ChatGPT
Use when proposing a new board before creating it: preview, mock up, or sketch what a board would look like as a table, with columns and example items. [UI COMPONENT] Renders an interactive table preview visualization that the user can see and interact with. IMPORTANT: This is a UI DISPLAY tool - use it to RENDER visual components for the user to see and interact with. Do NOT use data-fetching tools when the user explicitly asks to "show", "display", "visualize", or "see" something visually. Renders a MOCK-UP of a board that does NOT exist yet, from columns and example items you provide. Nothing is created in monday.com. Use it to let the user review a proposed board structure before you create it. After showing the preview, ask the user to confirm or adjust it, then create the board with the board creation tools only once they confirm. To show an EXISTING board, use show-table with its boardId instead.

show_connect_integration

ChatGPT
Show a widget to connect any monday integration (gmail, outlook, slack, jira) via OAuth. Render a widget with a button that opens a monday integration's OAuth consent flow (gmail, outlook, slack, jira, etc.) in a new browser tab, so the user can complete authorization directly rather than pasting a URL themselves. This tool cannot tell you whether the user completed consent; the widget reports it in the chat once it does. When a known integration ('gmail', 'calendar', 'outlook') already has an active connection, the widget shows that account as connected instead of a button, and the same setup a new connection gets runs right away.

stop_workflow_run_once

ChatGPT
Stops the run that run_workflow_once started, so the remaining steps do not execute. Use this tool only when the user asks to stop, cancel or abort the run. Stopping is best-effort: steps that already executed keep their effects — they are not undone. Say so rather than implying the run was reverted. The response is a JSON object: - status "ok": "stopped" tells you whether a run was actually stopped. When it is false there was nothing left to stop (the run already finished or never started), and "reason" may explain further. - status "failed": "message" explains why the stop request could not be made. Relay it.

submit_bug_or_feature_request

ChatGPT
Report a bug, submit a feature request, or share feedback about the monday.com product or this integration. Call this tool proactively — not just when a user explicitly asks. Use it whenever any of these signals show up: • A tool produced unexpected errors, empty results, or needed a workaround • The user tried something monday.com couldn't support and had to settle for a partial or manual solution • A recurring capability gap is noticed — something requested that simply isn't available in monday.com or this integration • The user shows repeated frustration (multiple corrections, retrying the same request, "that's wrong again," "why isn't this working") • A task required multiple retries, an unusually long reasoning chain, or many attempts for something that should've been simple Parameters: • title (string, required) — short summary, no PII • description (string, required) — full details of what happened/expected/requested, no PII • kind (enum, required) — "bug", "feature_request", or "feedback" • tool_name (string, optional) — the specific monday.com tool the feedback relates to (e.g. "create_item") Restriction: Use strictly for things related to monday.com — not for other tools (Google Drive, Slack, GitHub, etc.) that may be in the conversation context. Do NOT include any personally identifiable information (PII) such as names, email addresses, phone numbers, or any other personal data.

update_action

ChatGPT
Use to update fields of an existing saved action, such as its code, name, or variables. Update an existing action. Only pass the fields you want to change. Example: id: "550e8400-e29b-41d4-a716-446655440000", name: "Updated name", code: "print('new code')"

update_column

ChatGPT
Update properties of an existing monday.com column (title, description, settings). [REQUIRED PRECONDITION]: Uses optimistic concurrency control via the revision field — fetch the column id, type, and current revision via get_board_schema first (preferred), or get_board_info if you already have it, then call this tool. If the update fails because the revision is stale, re-fetch and try again. After a successful update, use the new revision returned in the response for any further update to this column, not the one you started with. [REQUIRED PRECONDITION]: If you are changing columnSettings, also call get_column_type_info with fetchMode "schema" for that column type first to learn the valid settings structure. columnSettings is the flat payload for that column type (e.g. {"labels": [...]}) — not get_board_info's column.settings object copied as-is, and not wrapped again as {"settings": {"labels": [...]}}. To edit existing status or dropdown labels (rename, recolor, or reorder): first call get_board_info with filters.columns.ids for that column (or filters.columns.only) to read its current settings.labels, where each existing label's id lives. Editing an existing label requires sending its id in that label's entry — omitting it fails validation, since only a brand-new label can omit id. Status labels need the full label shape (id, label, color, index, and so on), not just a renamed string. Never invent an id — reuse the ids from get_board_info and add new labels without one. Flow: get_board_info for the revision and current settings.labels with ids, get_column_type_info (schema mode) for the valid shape, then build columnSettings.labels reusing existing ids and adding new labels without id. To turn on or change the column rollup (calculated capability, for columns on multi-level boards), set rollupFunction: numbers MIN/MAX/SUM/NONE, date MIN/MAX, timeline MIN_MAX, status COUNT_KEYS. To turn rollup off, pass rollupFunction: null (omitting it leaves rollup unchanged). rollupFunction is independent of columnSettings and can be sent alone or together with it.

update_doc

ChatGPT
Update an existing monday.com document. Provide doc_id (preferred), object_id, or board_id + view_id, plus an ordered operations array (executed sequentially, stops on first failure). IDENTIFYING THE DOC: A doc can live in a workspace, on an item, or as a board view. For a doc that lives as a board view, you only have a board URL — pass board_id and view_id from https://<slug>.monday.com/boards/<board_id>/views/<view_id> and they will be resolved to the doc. OPERATIONS: - set_name: Rename the document. - add_markdown_content: Append markdown as blocks (or insert after a block). Best for text, headings, lists, simple tables — no block IDs needed. - update_block: Update content of an existing text, code, or list_item block in-place. - create_block: Create a new block at a precise position. Use parent_block_id to nest inside notice_box, table cell, or layout cell. - delete_blocks: Permanently delete 1–100 blocks in one call. Provide all block IDs in the block_ids array. The ONLY option for BOARD, WIDGET, DOC embed, and GIPHY blocks. - replace_block: Delete a block and create a new one in its place (use when update_block is not supported). - add_comment: Create a new comment or reply on the document (doc-level, block-level, or text-selection). WHEN TO USE EACH OPERATION: - text / code / list_item → update_block. Use replace_block to change subtype (e.g. NORMAL_TEXT→LARGE_TITLE) - divider / table / image / video / notice_box / layout → replace_block (properties immutable after creation) - BOARD / WIDGET / DOC / GIPHY → delete_blocks only GETTING BLOCK IDs: Call read_docs with include_blocks: true — returns id, type, position, and content per block. BLOCK CONTENT (delta_format): Array of insert ops. Last op MUST be {insert: {text: "\n"}}. - Plain: [{insert: {text: "Hello"}}, {insert: {text: "\n"}}] - Bold: [{insert: {text: "Hi"}, attributes: {bold: true}}, {insert: {text: "\n"}}] - Mention user/doc/board: [{insert: {text: "Hey "}}, {insert: {mention: {id: 12345, type: "USER"}}}, {insert: {text: "\n"}}] — type is USER, DOC, or BOARD. id is numeric (user IDs from list_users_and_teams) - Inline column value: [{insert: {column_value: {item_id: 111, column_id: "status"}}}, {insert: {text: "\n"}}] - Supported attributes: bold, italic, underline, strike, code, link, color, background (not applicable to mention/column_value ops) IMAGE WITH ASSET: For asset-based images, use create_block with block_type "image" and asset_id (instead of public_url). add_markdown_content does NOT support asset images — for mixed content, alternate add_markdown_content (text) and create_block (image) operations in sequence. BATCHING DELETES: delete_blocks accepts 1..100 IDs. Put ALL IDs in one operation's block_ids array. Never emit multiple delete_blocks operations in a row. COMMENTS: - add_comment: Create a new comment or reply on the document. Three scopes: - Doc-level (no block_id): comment appears on the doc as a whole. - Block-level (block_id only): comment is anchored to a specific block. The block shows a comment indicator in the UI. - Text-selection (block_id + selection_from + selection_length): comment is anchored to a specific character range inside a text/code/list_item block. That text is highlighted with a comment marker. Block-level and text-selection comments only work on blocks with text content (text, code, list_item, title, quote). They do NOT work on: divider, page_break, table, layout, notice_box, image, video, or giphy blocks. Get block IDs from read_docs with include_blocks: true. Format body with HTML, not markdown. Use mentions_list for @mentions.

update_folder

ChatGPT
Update an existing folder in monday.com

update_form

ChatGPT
Update a monday.com form. Use the action field to specify the operation. [REQUIRED PRECONDITION]: Call get_form first to read the current form state — you need it to resolve the formToken, and for actions that reference existing entities (updateQuestionOrder needs the question ids, deleteTag needs the tag id) or that overwrite existing settings (updateAppearance, updateAccessibility, updateFeatures, updateFormHeader).

update_items

ChatGPT
Update column values for up to 40 items in a single call. Each update targets one item by itemId and sets one or more column values on it. Each update is independent - it can target its own board via boardId and set its own column values, so a single call can update many items across multiple boards, apply the same value to many items, or apply different values per item. Each update returns its own item_id and item_url on success or a raw error message on failure. Always pass the board id, either once as the batch boardId or on each update. For subitems, pass the subitems board id. To link board-relation columns, call link_board_items_workflow before using this tool. Board-relation and dependency responses include linked item IDs up to 1000. Larger lists are omitted with a count and a message confirming the update succeeded. [REQUIRED PRECONDITION]: Before using this tool, if you are not familiar with the board structure (column IDs, column types, status labels), first use get_board_info with filters.columns.only to get column metadata without fetching views. This is essential for constructing valid column values.

update_view

ChatGPT
Update an existing board view (tab) — change its name, filter rules, or sort order. Provide only the fields you want to change. Filter operators: any_of, not_any_of, is_empty, is_not_empty, greater_than, lower_than, between, contains_text, not_contains_text

update_view_table

ChatGPT
Update an existing table-type board view — change its name, filters, sort, tags, or table-specific settings including conditional coloring (highlight rows/cells based on column values). Provide only the fields you want to change. CONDITIONAL COLORING: Use settings.conditional_coloring to highlight rows or cells. Each rule specifies a column_id, operator, value (human-readable — e.g. "Stuck", not an index), color, and entire_row flag. Example: highlight rows where Status is "Stuck" in red, or where Salary > 100000 in green. Filter operators: any_of, not_any_of, is_empty, is_not_empty, greater_than, lower_than, between, contains_text, not_contains_text

update_workspace

ChatGPT
Update an existing workspace in monday.com

validate_workflow

ChatGPT
Validates the current workflow's structure and step configuration. Reports issues such as a missing trigger or action block, a delay/wait-trigger block left as a leaf, an empty loop, unknown blocks, missing required inputs, type mismatches between a variable and the field it's bound to, cross-branch node-results references, invalid variable values, or switch steps whose cases are empty, have no matching route, or are missing entirely. Use this tool when: - The user asks "is my workflow ready?", "what's missing?", "can I publish?", "validate my workflow", or similar. - After you finished structural changes, to confirm the user still has things to configure. - Before suggesting the user publish/activate the workflow. The tool does NOT modify the workflow. It only inspects the current state. The response is always a JSON object with an "issues" array; an empty array means the workflow is fully configured. Each issue has a "code" discriminator with code-specific fields, and (when applicable) is enriched with stepNumber, stepTitle, and blockName for human-readable context.

vibe_ask

ChatGPT
Ask a read-only question about an existing Vibe app. Blocks for up to 45s (configurable via timeout_ms) awaiting the assistant reply. Status: COMPLETED with the reply, TIMEOUT if the workflow did not finish in time (call vibe_get later to retrieve it), or FAILED if the workflow errored or was cancelled. Optional model to pick the LLM for the answer.

vibe_collaborators

ChatGPT
Manage collaborators on a Vibe app. action=list returns all current collaborators. action=add grants a user editor access (user_id required; only the EDITOR role can be granted). action=remove revokes a user's access (user_id required). The app owner cannot be removed — use transfer_ownership instead. An editor may remove themselves to leave the app.

vibe_create

ChatGPT
Creates a new Vibe app from a natural-language prompt. Returns immediately with app_id and editor_link — the URL of the Vibe builder/chat page for the new app (https://{accountSlug}.monday.com/vibe/app/{appId}); the user can open it right away to watch generation in progress. Generation itself runs asynchronously — poll vibe_get for status. Optional: workspace_id to create the app in a specific workspace, board_ids to connect existing boards (omit to auto-create), view_id to host a dashboard widget, and model to pick the LLM.

vibe_delete

ChatGPT
Delete a Vibe app and its associated assets. Destructive.

vibe_duplicate

ChatGPT
Duplicate an existing Vibe app into a new app, copying the source app's live version. If workspace_id is omitted, the copy goes to the source app's workspace (or the user's default Vibe workspace). Returns the new app, including its editor_link. The copy is built in the background — poll vibe_get on the returned id until it is ready. Only apps that have a live (published) version can be duplicated, and not every app type supports duplication. To change an app in place, use vibe_update; to rename it, use vibe_rename.

vibe_get

ChatGPT
Fetch a Vibe app by id. App metadata is always returned, including editor_link — the URL of the Vibe builder/chat page for this app (https://{accountSlug}.monday.com/vibe/app/{appId}); usable as soon as the app row exists. Pass include to add expensive slices: status (refreshes status + adds is_busy, default true), messages (with optional from_date), code (generated files — the current version unless version_id or version_status is given), code_versions.

vibe_list

ChatGPT
List Vibe apps owned by the authenticated user. Supports pagination, search, status, and is_published filters.

vibe_publication

ChatGPT
Manage the publication state of a Vibe app on the caller account. action=publish requires the app to be deployed and respects the published-apps license limit. action=unpublish removes the app from the account. Returns app_id, success and message.

vibe_rename

ChatGPT
Rename a Vibe app and return the updated app.

vibe_update

ChatGPT
Sends a follow-up message to modify an existing app. Fire-and-forget — returns immediately with app_id, user_message_id and editor_link (the Vibe builder/chat URL for this app, https://{accountSlug}.monday.com/vibe/app/{appId}). Returns APP_BUSY (409) if the app is currently generating; poll vibe_get first. Optional model to pick the LLM for this build.

workspace_info

ChatGPT
This tool returns the boards, docs and folders in a workspace and which folder they are in. It returns up to 100 of each object type, if you receive 100 assume there are additional objects of that type in the workspace.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/monday-crm"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/monday-crm)