Cursor 1.5: Linear Integration, Smarter Agent Terminal, and Composer-Powered Workflows

Cursor 1.5, released in August 2025, bridges the gap between issue tracking and AI-assisted coding. With Linear integration, a smarter Agent terminal, and native OS notifications, it makes Composer—Cursor's frontier AI coding model—faster and more connected to your team's workflow. Here's what changed and why it matters for developers building with AI.

Run Agents Directly from Linear

The standout feature in Cursor 1.5 is the Linear integration. You can now start Background Agents directly from Linear without leaving your issue. Assign any issue to @Cursor or mention @Cursor in a comment, and Cursor will analyze the issue and kick off a background agent powered by Composer.

Cursor automatically pulls in issue details, comments, and linked references to give the agent the context it needs. You can create triaging rules that automatically assign certain tasks to the Cursor Agent—helping you respond immediately to user feedback, bugs, or feature requests. When the agent completes, team members can review diffs, pull requests, and even create PRs directly from Linear.

This closes the loop between project management and code generation. Instead of copying issue descriptions into Cursor manually, the integration keeps the agent and your team in sync from a single workflow.

Composer: The Engine Behind the Agent

Composer is Cursor's own AI coding model, introduced with Cursor 2.0 in October 2025. It is a mixture-of-experts (MoE) language model specialized for software engineering through reinforcement learning. On Cursor's internal benchmarks, Composer achieves frontier coding results with generation speed four times faster than similarly intelligent models.

Composer is trained with tools like codebase-wide semantic search, file editing, grepping, and terminal commands. It learns to make efficient choices in tool use and to maximize parallelism when possible. That makes it well-suited for the fast, interactive agent experience Cursor 1.5 now delivers, especially when triggered from Linear.

Most agent turns complete in under 30 seconds. For teams using the Linear integration, that means assigning an issue to Cursor can yield a pull request within minutes, without leaving your issue tracker.

Improved Review Flow in the Agent Terminal

Cursor 1.5 improves the Agent terminal experience. The terminal now opens on the left with a clear backdrop and border animation to highlight when it's blocking on your input. When you reject a command, the input field auto-focuses so you can respond immediately.

These changes reduce friction when reviewing agent actions. You spend less time hunting for the input field and more time deciding whether to approve or reject a command. The visual distinction between blocking and non-blocking states makes it easier to track when the agent is waiting on you.

OS Notifications from the Agent

Cursor 1.5 adds native OS notifications when an agent run finishes or when input is required—for example, approving a command that is not allowlisted. You can enable this from Settings.

This is especially useful when running background agents. If you delegate a task from Linear and switch to another window, you get a notification when the agent completes or needs your approval. No more checking back every few minutes to see if the run is done.

MCP Elicitation Support

Cursor 1.5 also supports MCP elicitation, a feature in the MCP (Model Context Protocol) spec that allows servers to request structured input from users. MCP servers can ask for user preferences or configuration choices, defined with JSON schemas. Users stay in control of what they share while servers get validated responses.

This gives MCP integrations more flexibility. A server can prompt for API keys, environment variables, or workflow preferences in a structured way, instead of relying on free-form prompts or pre-configured values.

What This Means for RAG and AI Pipelines

Cursor 1.5's tighter integration with Linear and its improved agent UX reflect a broader trend: AI coding tools are moving from isolated chat interfaces to connected workflows. For teams building RAG systems, agent pipelines, or knowledge-intensive apps, the pattern is similar: the value is in how the AI fits into existing tools and processes.

Platforms like ShinRAG take the same approach. Visual pipeline builders, embeddable widgets, and orchestration layers let you connect retrieval, synthesis, and integration nodes without writing orchestration code by hand. The goal is to make AI-powered workflows feel like a natural extension of your stack, not a separate tool.

Getting Started

To use the Linear integration, connect your Linear workspace from the Cursor dashboard. Once connected, assign issues to @Cursor or mention it in comments to start background agents. Enable OS notifications in Settings to stay updated when agents finish or need approval.

Cursor 1.5, together with Composer, shows how AI coding tools are evolving: faster, more integrated, and better aligned with how teams actually work.