We’re exploring Linear’s agent integration and evaluating whether to use Cursor’s built-in solution or build our own. The issue is that we can’t seem to customize the system prompt injected when a Background Agent is triggered from Linear; either via subassignment or via a @Cursor comment.
The hardcoded prompt is:
You've been assigned to resolve a Linear issue. Carefully analyze the issue and solve it by responding or writing code. Below you'll find the issue context and message history. Be sure to run and test code if it helps with your task.
We need to override this with a custom prompt (e.g., to instruct the agent to use specific MCP tools and scope its actions i.e. just triage the ticket and leave comments) . I couldn’t find any option for this in the settings page, the docs, or the agent configuration UI.
Is there any way to customize or override the Linear agent prompt? If not, consider this a feature request as in the ability to define a custom system prompt per integration source would make Background Agents significantly more useful for teams with established workflows.
Right now, the built-in Linear delegation flow, assigning to Cursor or using @Cursor comments, uses a hardcoded system prompt, and there’s no way to override it. The [key=value] syntax only supports repo, branch, and model.
Two things that might help, depending on your workflow:
AGENTS.md: Cloud Agents read AGENTS.md files from your repo. This won’t replace the hardcoded prompt, but it adds extra instructions the agent will follow. You can add triage-specific guidance, MCP tool usage instructions, and more there. Docs: Rules | Cursor Docs
Automations with Linear triggers: Automations support Linear triggers like “Issue created”, “Status changed”, and “End of cycle” and give you full control over the prompt, MCP tools, and scoping. The downside is these triggers are different from the native assign or @Cursor flow, since they fire on issue lifecycle events rather than delegation. If your workflow can adapt, this gives you the customization you’re looking for. Docs: Automations | Cursor Docs
A custom prompt per integration source is a reasonable ask, especially for teams that want agents scoped to specific actions like triage-only.
Let me know if either option works for your use case.
I agree that it would be great if we could update the hardcoded prompt. Usually, I start my tickets with an analysis or a plan. I tried a homemade integration with Asana and it worked well. I wanted to replace it with Linear, since it seems more integrated and requires less maintenance, but this hardcoded prompt breaks the workflow.
We can’t really reproduce the plan/ask workflow we use, like we can in the Cursor IDE.
Being able to override the prompt could really help.
