Feature Requests

It'd be dope if clickup AI would support https://modelcontextprotocol.io/ . With that its agent could instantly integrate with a ton of tools without you having to do first party integrations in the first place. Some tools like cursor.ai just allow you to paste in an MCP link and bam, you have it in your agent. Support this ONE thing and save time integrating all the others. Let your users do the job and give them the power to integrate properly.

I would like to suggest introducing a feature that allows exporting details (e.g., space, folder, list, etc.) related to the use of automations and AI. This would make it easier to locate the automations being used and quickly understand which ones have the greatest impact on costs. Currently, there are two separate functionalities that cannot be correlated and lack the ability to export or filter: Usage Details / Usage Breakdown with total numbers Log of executed automations Similar to: https://feedback.clickup.com/feature-requests/p/feature-usage-list

Since Clickup is deep in openai ecosystem, please upgrade the underlying model powering Clickup Brain's searching and writing with GPT-5 full version instead of 4o. Searching is still very bad with Brain and it's writing could be improved heavily.

Hi ClickUp Support Team, I’m using ClickUp Brain to audit and migrate a CRM based project pipeline into a new “Work In Progress” Space. The workflow is very close to being fully automatable, but the current set of actions Brain can execute is missing a few key capabilities that would unlock a complete end to end migration. 1) What ClickUp Brain was able to do successfully (in my workflow) Data review and reporting Cross-reference items from a “Backlogs” view against a CRM list (“Customers”) to find the matching customer card. Extract signals from customer cards (custom fields, tags, descriptions, notes) to infer “Products Ordered” and generate a structured table: Project Title Products Ordered (Website, LCRM, etc.) Link to CRM Customer Card Structure build (partial) Create Lists inside existing Folders (once the user manually created the folders first). Populate a “Client Info & Wiki” list and include a link to the CRM customer card in the List description. Create product lists following naming conventions (Website, LCRM, Podcast, etc.) once the naming rules were clarified. Validate that provided URLs corresponded to actual folders and read which lists already exist inside them. 2) Key limitations that blocked full execution (against the exact requests) A) Folder creation is not supported Request: “Create a folder per project in a target Space.” Limitation: Brain cannot create folders; only lists can be created. I had to manually create all project folders and provide folder URLs. Improvement request: Add an API/action for Brain to: Create folder in a specific Space Set folder color (optional) B) Cannot delete lists (cleanup is manual) Request: “Remove unnecessary lists like ones named ‘List’ and remove duplicates inside folders.” Limitation: Brain cannot delete lists. It can only identify what should be removed. Improvement request: Add actions for Brain to: Delete a list Bulk delete lists by filter (example: name equals “List”, or duplicate name within same folder) C) Cannot rename lists or update list metadata Request: Standardize list naming (example: convert “Website Build” and “LCRM Setup” into “Website” and “LCRM”). Limitation: Brain cannot rename lists or modify list settings/metadata. Improvement request: Add actions for Brain to: Rename lists Update list icon Update list color Update list description (Brain can set description on creation, but not reliably fix existing lists afterward) D) Cannot copy list icons/colors to match an existing template folder Request: “Use the same folder color and list icons I used for ‘Hot Blood Performance’.” Limitation: Brain cannot read or apply folder/list icon and color settings via the current tools. Improvement request: Add actions for Brain to: Read folder color and list icons from a reference folder Apply the same styling to newly created folders and lists (or allow a “clone style from template folder” operation) E) Limited ability to execute view based auditing without user provided item list Request: “Review all projects in a specific Backlogs view, then cross-check against CRM list.” Limitation: Brain struggled to directly pull a precise set of items from a specific view without me providing the list of items (via screenshot or explicit titles). The view itself was not a reliable source of item enumeration. Improvement request: Add support for Brain to: Retrieve the exact tasks/items shown in a specific view (including grouping and filtering results) Export the view’s visible task set as a structured list for processing 3) Outcome impact Because of the missing “folder create”, “list delete”, “list rename”, and “icon/color copy” actions, Brain could not complete the migration end to end. It could do the analysis and generate the target structure, and it could create lists once folders existed, but it could not: Create the folders Clean up placeholder lists (“List”) Remove duplicates Enforce styling consistency (icons/colors) 4) Summary of feature requests To enable full migration automation, please consider adding Brain supported actions for: Create folder in Space (and optionally set color) Delete list (single and bulk) Rename/update list metadata (name, icon, color, description) Read and clone styling from a reference folder/list (icons/colors) Retrieve the exact item set from a view (honoring filters, grouping) Happy to provide screenshots, workspace links, and a full reproduction path if helpful. Thank you, Liz

The pricing model is a miss in my opinion. Too many a la carte options, you're expecting an organization to mix and match and hope for the best. Not sure what adoption looks like, but it's confusing and not articulating the value clearly. Also, it feels like all or nothing, I need to purchase "AI" for my entire user base when I know only 20% would be getting value from it. Hard to perceive good value to the company when adoption is low. Your team should include all AI into the base product, so it's not an add on. Have all features available, then pull from a credit pool so the 20% of the users can access and increase the productivity. It'd be a ramp up vs all in. You'd be moving from a push to a pull style adoption where users are seeing wins from their colleagues. This increases ClickUp consumption and now direct value attached to it, which makes it a no brainer to add more to the credit pool. The name of the game is consumption, not seats :)

Introduce an Admin‑level control that allows Workspace Owners or Admins to enable, disable, or restrict access to specific LLM models used by ClickUp AI. This would be surfaced as part of Workspace Settings → AI / ClickApps / Security & Permissions. Suggested Capabilities Model Visibility Controls Allow admins to see which AI/LLM models are available in their workspace (without exposing them to end users by default). Model Enable/Disable Toggles Enable or block specific LLMs (e.g. ChatGPT, Claude, Gemini). Role‑Based Restrictions Allow different models (or none) for different roles (Admins, Members, Guests, Limited Members). Compliance Mode Option to restrict AI usage to a single approved model for regulatory or audit purposes. Audit & Transparency Logging/reporting indicating which model category was used for AI‑generated outputs (admin‑visible only).

Treat .tmdl as a text file and surface its contents to AI and search so they can be uploaded to comments/tasks and indexed when a folder that contains tdml files is shared (via OneDrive, GoogleDrive, GitHub etc.)

It would be beneficial if the AI Notetaker had better integration with Microsoft Teams. Our organization uses Teams for their virtual meetings. Because the ClickUp AI Notetaker is not authenticated with our domain it is considered an external user. Microsoft Teams puts the Notetaker into the lobby and someone has to accept it into the meeting. Once accepted into the meeting it appears like another person in the meeting. This is very confusing to anyone in our organization who doesn't use ClickUp and it just seems strange to have a notetaker represented as a person in the meeting. Asks: Please develop some way that the AI Notetaker can authenticate so it doesn't end up in the lobby. Please develop a cleaner approach to having the Notetaker take notes during meetings instead of being represented as another person in the meeting.

The get_chat_channels tool returns DM channels with only type , visibility , creator , and created_at — there is no members or counterparty field. For an agent, this makes it impossible to identify which DM channel corresponds to which user without probing messages in every channel. Repro: 1. Authenticate against mcp.clickup.com/mcp as a user with several DMs 2. Call clickup_get_chat_channels 3. All DM entries list the viewer as creator with no indication of the other party Impact: - An agent cannot send a DM to a named user without enumerating every DM channel and checking its messages for a user_id match. - Today's workaround: the user manually seeds a DM in the ClickUp UI, then the agent probes each DM's messages looking for the one with fresh activity. - Scales poorly (O(n) API calls per user lookup), and is brittle on empty or cold DMs. Suggested fixes (pick any): 1. Include members: [user_id, …] (or an explicit counterparty_user_id for DMs) in the get_chat_channels response. 2. Add a clickup_find_or_create_dm_channel tool that takes a user ID and returns the DM channel ID (creating it if missing). 3. Allow clickup_send_chat_message to accept a recipient_user_id alongside (or instead of) channel_id , auto-resolving/creating the DM. This mirrors Slack's chat.postMessage , which accepts a user ID as channel and auto-opens the DM. Context: Hit while integrating the first-party mcp.clickup.com/mcp server into an internal MCP Manager. Happy to test any proposed fix.

DOCX is currently not allowed as a format to upload to the clickup chat. It would be good to be able to upload docx-like word documents, not just txt and pdf