AI Category

Hi there, I am currently working on a use case that would involve retrieving all information from boards/projects using MCP, so that the agent can extract relevant information needed.I have familiarised myself with all the MCP tools from here and I have created an AI agent that can use these tools, however, I have noticed that whenever agent is using tools get_board_items_page and get_board_info to extract all relevant information available in the board it doesn't retrieve any subitems or linked columns, when I specifically name the subitem name or linked column name, it simply comes back to me saying it can't find it. The payload returned from using these tools confirms is as well. Could you please help me understand if I am missing anything? Is there another tool that should be used for linked columns and subitems? I have noticed that there is a tool create_item which allows the creation of subitem, so does that mean the current MCP tools don't allow to read subitem and linked column

Hey everyone!If you use Claude Code, you can now install a plugin that gives it actual knowledge of the monday code development lifecycle. No more pasting documentation, no more explaining what .mondaycoderc is or how to use logs in production. Introducing the agentic monday apps framework - a plugin marketplace for Claude Code, built for monday app developers with monday-code as the first plugin. Five slash commands, full SDLC/monday-code-init scaffolds a new app — frontend (Vite + React + MondayContext), backend (Express + JWT + Document DB), or both. Multi-tenant isolation patternincluded. Vibe Design System wired up. You run one command and get a working codebase./monday-code-migrate is for existing apps. It reads your project first, shows you what it plans to change, and waits for your OK before touching anything. It adds monday-code compatibility without rewriting your business logic./monday-code-dev handles local setup: dependency checks, local MongoDB via Docker if you're using

Hey Community,Last week marked a massive shift in how we think about work. We’ve just unveiled that monday.com is now officially open to AI agents. This isn't just a new feature; it’s the beginning of a new era for the monday.com Marketplace. We believe the future of work is a seamless collaboration between humans and agents, and we want our partner ecosystem to be at the very forefront of this evolution. This is just the beginningWe have much more to share with you in the next few days. We are rolling out more tools, deeper documentation, and new ways for you to monetize and distribute AI-driven solutions.Stay tuned for a major announcement coming soon.In the meantime, check out the new Agent Experience and start imagining how your apps will power the next generation of autonomous work.Why this matters for youWe’ve opened our infrastructure to make monday the ultimate "brain" for AI. From our new HATCHA (reverse CAPTCHA) verification that lets agents in, to native MCP (Model Context P

If you're using Cursor and want to connect the monday.com MCP servers (both the platform API and the apps framework), here's a guide covering the real-world problems you'll run into and how to solve them. This is especially relevant if you work with multiple monday.com instances (e.g., dev, staging, production, or separate accounts for different purposes). The setupmonday.com provides two MCP packages: @mondaydotcomorg/monday-api-mcp — for workspace automation (platform API)@mondaydotcomorg/monday-api-mcp@latest --mode apps — for the apps frameworkBoth are configured in ~/.cursor/mcp.json. For a single instance, the docs work fine. But things get interesting when you have multiple instances. Problem 1: The ENOTEMPTY race conditionWhen you configure multiple MCP servers using the same npx package, Cursor starts them all simultaneously at launch. Each one tries to install/update the package into the same shared npx cache directory — and they collide:npm error ENOTEMPTY: directory not emp

Hey Community! For those of you who don’t know, I’m not a technical person, but we've had our Apps MCP for about 3 months, so I decided to challenge myself by building a simple to-do list application for monday with it.You can check out my journey, which documents: The step-by-step process of using monday.com Apps MCP to construct a functional tool. Creating a solution with task lists and checkboxes to manage items like cooking dinner or cleaning the house. And yes - it took some “prompt fixes” from my side, but in less than 30 minutes (less than 22 minutes, originally) the app was up! Check out the full video linked below for the complete build walkthrough and share that excitement!Let me know in the comments: what’s one small, useful app you'd love to build for yourself or your team without needing to code? Or - what would you add to your existing app/s leveraging the Apps MCP? 

A while back, I accidentally sent the same Slack message five times to hundreds of community membersIt happened because I was relying on a fragile automation flow. One small mistake and it cascaded.Instead of trying to patch the automation again, I decided to rebuild the workflow from scratch using AI.I used Cursor to generate a small internal tool that: Connects as a Slack bot Lets me send controlled bulk messages Adds simple guardrails to prevent duplicate sends Includes a preview step before pushing anything live Logs activity so I can trace what happened Time to working version: about 30 minutes. It’s not a polished SaaS product. It’s a focused internal tool built around one very specific pain point.What stood out to me is how different the mindset feels now. A year ago, I would have searched for a tool or tried to stretch an automation platform beyond its limits. Now, if the logic is clear, I can just build the layer I need.For those of you building products: Are you using

Over the past few days, the internet has been buzzing about a new open-source AI assistant called Open Claw.For anyone who has not followed it yet, Open Claw is an open-source AI agent released late last year that can write code, edit files, and browse the web to complete tasks on behalf of users.But there is another side effect that no one really expected:a sudden race to buy Mac Minis. A computer that starts at around $600.Naturally, I wanted to try it myself.TL;DR: it was not easy, but it was genuinely fun.A few things upfrontBefore jumping in, there were a few important constraints and thoughts:I did not want to run this on my personal computer and definitely not on my work laptop. Security: at the end of the day, this means giving an open-source project very broad access to a machine. You really need to understand the risk and what you are comfortable with. Open Claw agents seem to have a life of their own. Just a few days ago they opened their own social network where they compla