Elle, the AI assistant in Aha! software, can help you design, create, and refine product work within Aha! products. When you want to access Aha! data within another AI tool like ChatGPT, Claude, or Cursor, you can connect to the remote Model Context Protocol (MCP) server.
From your AI tool of choice, you can use the MCP server to read and write Aha! data. Your AI tool will be able to search for records, summarize work, create new records, and edit existing ones, all through natural language prompts.
Click any of the following links to skip ahead:
Confirm user permissions
The AI tool accesses the MCP server with the same level of permissions your user has in your Aha! account, and uses Elle, the Aha! AI assistant, to complete some tasks. You must have AI functionality enabled in your Aha! account for the MCP server connection to work.
Action |
User permissions |
|---|---|
Connect to and use the MCP server |
Contributor |
An account administrator can enable or disable read and write access to the MCP server separately from User menu -> Settings -> Account -> AI controls.
Connect to the MCP server
For most AI tools, connection is as simple as adding the following server URL in your AI tool's MCP settings:
https://[your-aha-domain].aha.io/api/v1/mcp
For setup instructions specific to your tool, refer to its MCP documentation. Most tools ask for a Name and a Server URL. Use "Aha!" as the Name.
For example, here is how you would connect to Claude.ai. From Claude:
Navigate to Settings, then click Connectors.
Click Add custom connector.
Enter "Aha!" as the name and paste the server URL above.
Click Add.
Once connected, query the MCP server using natural language prompts. The AI tool will be able to read, create, edit, analyze, and use as context data from your Aha! account.
The MCP server has the same rate limit as the Aha! REST API: 300 requests per minute, or 20 requests per second.
The rate limits are per IP address and per Aha! account. If you work on the same IP address as other members of your Aha! account, you may hit the rate limit sooner than otherwise.
Read and write Aha! data
The remote MCP server gives your AI tool direct access to your Aha! account data. What you can do depends on your user permissions and the access your account administrator has enabled.
Read: Search for records, retrieve record details, fetch report data, and summarize work across workspaces and teams.
Write: Create new records, edit existing records, add comments, copy records, and create record links.
The MCP server cannot delete records, but it can clear, edit, and overwrite field values.
All actions are subject to the same permissions your user has in Aha! software. If you cannot edit a record in the Aha! interface, you cannot edit it through the MCP server.
Changes made through the MCP server appear under your name in audit logs and user activity. This is the same as if you had performed the action directly in Aha! software.
Your account administrator can control MCP server access from User menu -> Settings -> Account -> AI controls. Read access and write access are separate settings, so your organization can enable read-only access if it prefers to restrict editing to the Aha! interface.
AI credits
Your external AI tool's processing does not consume Aha! AI credits. Straightforward data retrieval from your Aha! account and record updates to your Aha! account do not consume Aha! AI credits either. However, if a request requires Aha! to use AI to interpret, summarize, or generate a response, that request will consume Aha! AI credits.
Most MCP requests are straightforward data retrieval or record updates that do not invoke Aha! AI. These include:
Reading or retrieving records
Finding workspaces and teams
Fetching report data
Adding comments
Creating or editing records
Copying records
Creating record links
Some MCP requests do use Aha! AI internally:
Searching records: Aha! AI may filter search results for relevance.
Analyzing records: Aha! AI summarizes or synthesizes information across multiple records.
Setting up reports: Aha! AI narrows a large set of fields to the ones relevant to the requested report.
The remote MCP server does not consume AI credits for every request. Credits are consumed only when a request requires Aha! AI to process internally. For these AI-assisted requests, credit usage varies based on how many records or fields need to be processed, how large the record content or report metadata is, and whether the request requires one or multiple AI passes to complete. There is no fixed credit cost per MCP connection, request, or action — usage is variable.
Example queries
Use these as a starting point. Insert your own record names, releases, or topics.
Find what is already tracked
"Do we have anything tracking [topic]?"
"Search for ideas related to [feature area]. How many votes do they have?"
"Are there any open bugs related to [topic]?"
Check status
"What is the status of [record reference number]?"
"What features are assigned to me this sprint?"
"What is in the [release name] release and where does each feature stand?"
Summarize work
"Summarize all features that changed status this week."
"Give me a summary of the [release name] release before my planning meeting."
"What work is in progress across the [workspace name] workspace?"
Create and edit records
"Create a new feature in the [workspace name] workspace called [feature name]."
"Update the description of [record reference number] to include [details]."
"Add a comment to [record reference number] summarizing [topic]."
"Copy [record reference number] into a new record."
If you get stuck, please reach out to our Customer Success team. Our team is made up entirely of product experts and responds fast.