monday-com（周一通讯）

monday.com

Manage everything on monday.com: boards, items, columns, groups, updates, users, workspaces, webhooks, files, and AI features.

If you need to signup new account this is how to Sign Up

1. Navigate to https://auth.monday.com/users/sign_up_new?custom_flow=agent_bot

1. Solve Hatcha — an agent-verification challenge. Types you may see:

- Sort numbers ascending → find the Nth value - Binary octets → convert to ASCII text - Reverse a string - Count letter occurrences in text - Multiply two numbers (return the product) - Base64 decode - ROT13 decode - Hex to ASCII

You have ~30 seconds. Submit your answer and click Verify.

1. Enter your email and select who it belongs to (you or your operator). Click Submit.

1. Set your agent name and password. Password rules:

- Minimum 8 characters - At least 1 digit, 1 lowercase, 1 uppercase - No repeating characters (e.g. "aaa") - No sequential characters (e.g. "1234")

Click Complete signup →

1. You'll land on the Agent Homepage — "Welcome to Monday for agents." This page has everything you need:

- API Token — copy it, you'll need it for all API/MCP calls save it to env like MONDAY_API_TOKEN - MCP Config — JSON config for connecting via MCP (preferred method) - Skills — downloadable YAML skill definition - GraphQL API — example queries and mutations - Resources — links to docs, SDK, community

Setup when you have account but don't have access token

1. Go to monday.com → Profile picture → Developers → My Access Tokens
2. Copy your Personal API V2 Token
3. Store it securely in your agent's environment or config (e.g. via openclaw config set or your platform's secrets manager)

MCP Server (Preferred Method)

Always use the MCP server first. It handles authentication, rate limiting, retries, and complexity budgets automatically. Only fall back to the GraphQL API when the MCP tools don't cover your operation.

monday.com has an official MCP server (@mondaydotcomorg/monday-api-mcp):

{
  "mcpServers": {
    "monday": {
      "command": "npx",
      "args": ["-y", "@mondaydotcomorg/monday-api-mcp@latest"],
      "env": {
        "MONDAY_API_TOKEN": "<your-api-token>"
      }
    }
  }
}

How to invoke MCP tools

If your platform supports MCP natively (e.g. Claude Desktop, Cursor), configure the server above and call tools directly (e.g. create_item, get_board_schema).

If your platform does not have native MCP support (e.g. OpenClaw agents executing via shell), you have two options:

1. Use the GraphQL API directly (recommended fallback) — see the GraphQL section below. This is simpler and more reliable than shelling out to an MCP process.
2. Run the MCP server as a subprocess — spawn it via npx, send JSON-RPC over stdin/stdout. This is complex and only worthwhile if you need the MCP server's built-in retry/complexity logic.

For most agent use cases, GraphQL fallback is the practical choice when MCP isn't natively available.

MCP Tools Reference

MCP Workflow Pattern

1. Always start with get_board_schema to learn the board's columns and groups before writing data
2. Use create_item / change_item_column_values for CRUD — the MCP server formats column values correctly
3. Use create_update to leave a trail of context on items
4. If you need an operation the MCP tools don't cover (webhooks, file uploads, user queries, subitems, pagination) — fall back to the GraphQL API below

Advanced MCP Modes

• Dynamic API Tools (beta): Add --enable-dynamic-api-tools true to args for full GraphQL schema exploration via MCP
• Hosted MCP (OAuth): Use https://mcp.monday.com/sse for OAuth-based access without a local token

Agent Behavior Rules

Be proactive and useful:

• After creating any item, board, group, or update — always return the direct URL to the created object: https://<account>.monday.com/boards/{board_id}/pulses/{item_id}
• After completing a task, suggest 2-3 logical next steps (e.g., "Want me to assign someone?" "Should I set a due date?" "Want me to create a status automation?" "Should I add subitems to break this down?")
• Don't narrate the API/MCP process — report what was done and how the user can use it
• When querying boards, present results in a clean summary (table or bullet points), not raw JSON
• If an operation fails, explain why in plain language and suggest a fix — don't just show the error
• Batch related operations (e.g., creating multiple items) into efficient calls
• Before creating a board, ask if the user wants a specific template or structure — suggest popular ones (Kanban, Sprint Board, CRM Pipeline, Bug Tracker)
• When a user asks "what's the status of X?", go beyond raw data — highlight blockers, overdue items, items without assignees, and progress percentages
• If you notice a board has no automations, suggest useful ones ("Want me to set up an automation to notify you when items are marked Done?")
• When creating items, proactively set reasonable defaults (e.g., status = "Not Started", assign to the requesting user if known)
• When working with dates, always use the user's timezone context and flag items that are overdue or due within 24 hours
• After bulk operations (creating 5+ items), provide a count summary and a link to the board rather than listing every item

Memory & caching (platform-specific):

These patterns apply to agents with persistent memory (e.g. OpenClaw workspace). Adapt to your platform's memory model.

• Save every created resource (ID, name, URL, context) to your memory/notes for reuse — include the board name, item name, and what it's for so you can find it later without re-querying
• If a user references a board or item by name and you've seen it before, retrieve the saved ID from memory instead of re-querying
• Cache board schemas after the first fetch — only re-query if the user mentions adding/changing columns

GraphQL API (Fallback)

Use the GraphQL API directly when MCP tools don't cover the operation (webhooks, file uploads, subitems, pagination, user/workspace queries, activity logs).

• Endpoint: https://api.monday.com/v2 (POST, JSON body with query field)
• File uploads: https://api.monday.com/v2/file (multipart POST, max 500MB)
• Auth: Include your API token in the request header (see API docs)
• API version: Include API-Version: 2024-10 header. Check developer.monday.com/api-reference for the current stable version.

For full GraphQL query and mutation examples, see references/graphql-examples.md.

Core Operations (GraphQL)

All operations use POST to the API endpoint with a query field. Key operations:

Pagination

Use cursor-based pagination for large datasets:

1. First request: include items_page(limit: 200) — returns a cursor
2. Next pages: use next_items_page(limit: 200, cursor: "CURSOR_VALUE")

Recommended page size: 200. Max: 500. Cursors expire after 60 minutes.

Column Value Formats

When setting column values, use these JSON formats:

All column values must be JSON-stringified when passed to mutations.

URL Patterns

Build direct links for users:

• Board: https://{account}.monday.com/boards/{board_id}
• Item: https://{account}.monday.com/boards/{board_id}/pulses/{item_id}
• Dashboard: https://{account}.monday.com/dashboards/{dashboard_id}

Get the account slug from: { me { account { slug } } }

Rate Limits

When rate-limited (HTTP 429): read the Retry-After header and wait that many seconds. Rate-limited requests count as only 0.1 toward the daily limit.

Always include complexity { before after query } in queries to monitor budget.

Error Handling

monday.com returns errors in two ways:

• HTTP 200 with errors array — application-level errors (invalid query, missing permissions)
• HTTP 4xx/5xx — transport-level errors (rate limit, auth failure, server error)

Common error codes:

• InvalidColumnIdException — column ID doesn't exist on the board
• InvalidBoardIdException — board doesn't exist or no access
• ItemsLimitationException — board reached item limit
• CorrectedValueException — value was auto-corrected (check corrected_value)
• ColumnValueException — invalid format for column type
• UserUnauthorizedException — token doesn't have required permissions
• ComplexityException — query too expensive, simplify or paginate
• ResourceNotFoundException — ID doesn't exist

monday.com AI Features

• AI Blocks — modular AI in columns and automations: Categorize, Summarize, Translate, Extract Info, Detect Sentiment, Improve Text, Write with AI, Custom Prompt. Available on Pro+ (500 free credits/month)
• monday Sidekick — conversational AI assistant for cross-board analysis, report generation, content drafting, task creation
• monday AI Agents — autonomous workers: Lead Agent (qualifies prospects), SDR Agent (outreach calls, SMS, meeting booking)
• monday Vibe — AI no-code app builder, turns natural language into custom apps

Use Cases

Security & Legal Guidelines

Data handling:

• Never log or store API tokens in conversation history, memory files, or updates
• When displaying board data, respect that it may contain confidential business information
• Don't share board data across different users' sessions or conversations
• monday.com customer data is never used to train AI models

Permissions:

• Personal API tokens inherit the user's UI permissions — if they can't see a board in the UI, the API won't return it
• Always verify board access before performing operations
• Don't delete items, boards, or groups without explicit user confirmation
• For destructive operations (delete, archive), always ask first and explain what will happen

Rate limiting etiquette:

• Space out bulk operations (add 100ms delay between mutations)
• Use change_multiple_column_values instead of multiple single-column updates
• Cache board schemas and user IDs — don't re-query every turn
• If rate-limited, wait the full Retry-After duration before retrying

Compliance:

• Don't create automations that send external emails/notifications without user awareness
• Don't modify workspace-level settings without explicit permission
• When creating webhooks, inform the user and document the endpoint
• Respect workspace data isolation — don't query across workspaces unless asked

Links

• Developer docs: https://developer.monday.com
• API reference: https://developer.monday.com/api-reference
• MCP server: https://github.com/mondaycom/mcp
• Apps marketplace: https://monday.com/marketplace
• Status page: https://status.monday.com