AI Super Agents

Agent tool - list-update/rename capability

Today, AI Agents can read relationships between tasks, but can't create, edit, or delete relationships between tasks. I would like an Agent to be able to automatically manage task relationships (task relationships, related relationships — not dependencies) based on the value of a custom field. Concrete example of use: I have a custom Project Id field on a list. When the Project ID of a task is changed, I would like an Agent to: find all the other tasks (and subtasks) that have the same Project ID, automatically adds a related relationship between the triggering task and these tasks, removes existing relationships to tasks whose Project Id no longer matches, without changing any other fields (status, assignee, etc.). This would make it possible to: automatically keep relationships between tasks in the same project synchronized, avoid manual errors or missing links, build more advanced workflows around the structure of projects (navigation by relationships, reports, linked views, etc.). Functionally, the Agent would need: read the value of a custom field on the triggering task, search for other tasks sharing the same value in a given scope (list/folder/space), create/delete “task relationship” relationships between tasks. This type of capability would open up a lot of advanced automation scenarios (link by customer, by campaign, by product, by sprint, etc.) based on custom fields.

We've been trying to use research agents for automation and direct conversations. However, we are consistently running into an issue where tool fails. We've tried all of the following and they all fail Super agent with web search enabled DM "Every..." automation with super agent and web search enabled "Every..." automation with "Anything with AI" with web search enabled The only time we are able to actually use web search is when chatting directly with Brain2.

Two things: It'd be great the possibility to limit how many credits each user can use per month per Super Agent It'd be great the possibility to know how many credits each user is using per Super Agent per day, week, month Why I need that ? I need to avoid one user to spend all IA Agent credits making excessive questions and be aware of that.

As a workspace with a growing number of custom Super Agents, we'd like the ability to organize them into user-defined categories or folders within the AI Hub. Currently all self-created agents appear in a flat list, which makes it harder to find the right agent as the number grows. Suggested solution: Allow users to create custom folders/categories in the AI Hub (similar to how Spaces organize work), or at minimum support tagging agents so they can be filtered.

Current Limitation: Super Agents are currently powered by models that may not meet everyone's needs in terms of performance, accuracy, or specific use case requirements. This one-size-fits-all approach limits the potential of AI automation within ClickUp. Proposed Solution: Add a "Bring Your Own LLM" feature that allows users to connect their own AI model API keys (OpenAI, Anthropic, Gemini, etc.) and choose which model powers their Super Agents so users maintain full control over model selection and costs Key Benefits: Enhanced Performance Use cutting-edge models (latest ChatGPT, Claude, etc.) for superior accuracy Select models optimized for your specific industry or use cases Access to specialized or fine-tuned models Cost Control & Flexibility Choose budget-friendly models for simple tasks Use premium models only where needed Manage AI costs directly through your preferred provider Pay for LLM tokens instead of ClickUp credits Customization Options Switch between models based on task complexity Use different models for different agents

I would like to request the ability for ClickUp AI Super Agents and Automations to apply different types of ClickUp templates automatically. Currently, it would be very useful if agents and automations could perform actions such as: Apply a Task Template Apply a List Template Apply a View Template Potentially apply other ClickUp template types supported by the workspace The main goal is to allow AI agents and automations to build standardized structures based on predefined company templates, instead of requiring users to manually apply them after the automation runs. For example, an AI Super Agent could: Create a new List for a specific project or collection. Apply the appropriate List Template. Apply one or more predefined View Templates. Create tasks using specific Task Templates. Continue managing the workflow based on the established automation rules. This feature would significantly improve standardization, scalability, governance, and automation of complex workflows, especially for organizations that rely heavily on ClickUp templates to maintain consistent processes. Ideally, these template actions should be available both as Automation actions and as tools/actions that AI Super Agents can execute.

Super Agents cannot grant themselves or other agents/users access to ClickUp documents programmatically. This blocks automated workflows where an agent creates or edits a document that other agents or users need to access. ## Current Behavior An agent (e.g., "Minutes Maven") creates an ATR document as a subpage inside a meeting notes doc. A second agent (e.g., "Status Sync Sam") needs to edit that document when task statuses change. The second agent fails with an access error because it does not have permissions on the doc. Adding a doc to an agent's workspace knowledge does not grant write access to private/restricted documents. @mentioning an agent in a doc comment notifies it but does not grant access. The only workaround is for a human user to manually open Share settings and add the agent as a collaborator. ## Expected Behavior At least one of the following: A "share_document" tool available to agents that allows them to add users/agents as collaborators (with a specified permission level) on a document they own or have edit access to. Automatic access inheritance : When Agent A creates a document page, other agents listed in Agent A's prompt or workspace knowledge should automatically receive edit access. Workspace knowledge = edit access : If a document is added to an agent's workspace knowledge, the agent should have read+write access, not just read/focus. ## Impact Multi-agent workflows are broken : Any workflow where Agent A creates a doc and Agent B later updates it requires manual human intervention to grant permissions every single time. Defeats automation purpose : The whole point of the agent is to eliminate manual steps. Having to manually share every new ATR doc with the sync agent adds friction that scales linearly with meeting volume. List permissions work differently : Agents can create tasks in any list added to their workspace knowledge without manual sharing. Documents should behave consistently. ## Use Case Meeting ATR Automation (Balasore Alloys / Preflex Solutions) AI Note Taker records a meeting → creates meeting notes doc (private to the note taker owner). "Minutes Maven" agent is triggered → reads the notes, creates an ATR subpage + ClickUp tasks with ATR doc references. HODs work on tasks, update statuses. "Status Sync Sam" agent fires on task status change → needs to update the ATR doc's status column. Step 4 fails because Sam has no access to the doc, even though it's in Sam's workspace knowledge. Volume : 20+ HODs, daily meetings = 20+ ATR docs/day that each need manual sharing. Unscalable. ## Suggested Solution Priority Option 1 (share_document tool) is the most flexible and aligns with how list permissions already work for agents. It would also solve the attendee-sharing use case (auto-share ATR with all meeting attendees). ## Environment ClickUp Enterprise plan Super Agents (Brain) AI Note Taker integration Multi-agent workflow (creator agent + updater agent)

As there is no current automation to mark a task as private, having an AI agent that can do this in the interim would be helpful. I would like to be able to mark a task as Confidential (checkbox = checked) and the task is automatically marked as private (ideally shared with a set group of people). This is for people management, e.g. an employee is being performance managed. A phone call or email to them is noted and assigned to their employee profile task.

I gave my superagent the skill to do websearch. When creating the AI it says it can search new contacts form the web but when I try to use it, it always says it does not have the skill.