Cursor introduced Cursor Router, an intelligent model routing system that powers Auto mode. It analyzes each request and directs it to the most appropriate model based on three optimization modes: Intelligence, Balance, and Cost. The router is available across desktop, web, iOS, CLI, and SDK, and is on by default for Teams plans.

Cursor in Slack now shares a plan before starting work and updates its status as it progresses. It can operate in multi-repo environments, switching between repositories mid-task as needed. Additionally, Cursor can now read from and send messages across different Slack channels and threads for cross-channel workflows.

Cursor introduced side chats that run alongside main agent conversations, allowing users to explore ideas without interrupting their primary workflow. Conversation search was added to find past agent chats by searching transcripts via the command palette or within conversations. Additionally, redesigned project and repo pickers and new cloud agent hooks for observing and controlling agent conversations were introduced.

Cursor admins can now configure Team MCP servers once and distribute them across cloud agents, the agents window, IDE, and CLI through team marketplaces, eliminating per-member server configuration. Team marketplaces also now support organization groups for access control, in addition to existing SCIM directory groups, enabling more granular enterprise integration governance.

Cursor launched a public beta iOS mobile app available on all paid plans, enabling users to launch and manage always-on cloud agents from anywhere. The app supports voice input, slash commands, Remote Control of desktop agents, Live Activities, push notifications, and the ability to review artifacts and merge PRs directly from the phone.

Cursor introduced a new Customize page that consolidates plugins, skills, MCPs, subagents, rules, commands, and hooks into one place, manageable at user, team, or workspace level. New features include a marketplace leaderboard showing popular plugins across teams, plugin canvases (prebuilt shared setup templates like Hex and Atlassian canvases), and team marketplaces now support imports from GitLab, BitBucket, and Azure DevOps.

This release introduces the /automate skill for creating automations directly in local agent sessions, new GitHub triggers (issue comments, PR review comments, PR review submitted, review thread updates, workflow run completed), and an emoji-based Slack trigger. Additionally, cloud agents kicked off by automations can now use computer use tools to produce demos or artifacts of their work.

Cursor now supports cloud environment setup in under 10 minutes with reusable snapshots for faster future agent startups. Cloud subagents can be spun up via /in-cloud to work in isolated VMs on separate branches, enabling parallel and long-running tasks. Improved handoff between local and cloud sessions allows offloading work and running multiple cloud agents in parallel.

Bugbot is now over 3x faster with average review time reduced to ~90 seconds from ~5 minutes, finds 10% more bugs per review, and costs ~22% less per run. These improvements are powered by Composer 2.5 which now powers Bugbot. A new /review command allows running Bugbot and Security Review before pushing code, with support for syncing with GitHub and GitLab PRs.

Cursor's Design Mode in the browser received updates including multi-select elements functionality, allowing users to click multiple UI elements simultaneously for batch editing. Voice input was also added, enabling users to narrate changes through the Design Mode overlay while agents are mid-run, allowing queuing of next changes without waiting.

Cursor introduced Design Mode for canvases, allowing users to select and annotate UI elements directly to guide edits more quickly. A new Context Usage Report feature was added, showing an interactive breakdown of token usage across system prompt, tool definitions, rules, and skills within a canvas. Users can also ask the agent follow-up questions to customize the report and identify opportunities to reduce context usage.

Cursor Enterprise now supports a three-tier hierarchy (Organizations → Teams → Groups) allowing companies to manage multiple teams from a single top-level organization with unified spend/token reporting. Teams can have independent security, governance, budget, and feature settings, while Groups enable fine-grained model access, spend limits, and agent permissions for user cohorts without requiring separate teams. This is now generally available to all Enterprise customers.

Cursor introduced Auto-review, a new run mode that allows the agent to work longer with fewer approval prompts and safer execution. It applies to Shell, MCP, and Fetch tool calls, with allowlisted calls running immediately and others going through a classifier subagent. The classifier subagent decides whether to allow, try a different approach, or request user approval.

Cursor Automations are now available in the Agents Window, allowing users to create and manage automations alongside their agents. The update introduces multi-repo automations for cross-codebase work and no-repo automations for non-code tasks. Five new automation templates were added to the Cursor Marketplace, including Slack digest, product analytics, product FAQ, product finance, and customer health agents.

Cursor introduced Shared Canvases, allowing users to share interactive artifacts created by agents (such as reports, dashboards, and custom interfaces) with their team via a link. Teammates can open a live snapshot of the canvas in the browser without needing to share the full chat thread. Shared canvases are viewable in the Cursor Dashboard with read-only access and are available on Pro, Teams, and Enterprise plans.

Cursor is now available as an integration in Jira, allowing users to assign work items to Cursor or mention @Cursor in comments to kick off a cloud agent. The agent uses work item details and repository settings to scope tasks like fixing bugs, adding features, or updating tests. Upon completion, Jira shows updates and includes a link to the generated pull request.

Composer 2.5 is a substantial improvement in intelligence and behavior over Composer 2, featuring better performance on sustained long-running tasks and more reliable execution of complex instructions. It is also described as more pleasant to collaborate with, with benchmark results showing significant improvements.

Cursor introduced new tools for configuring development environments for cloud agents, including multi-repo environment support, Dockerfile-based environment configuration with build secrets and improved layer caching (70% faster builds), and improved agent-led environment setup with validation and fallback handling. Additionally, environment governance and security controls were added, including version history, rollback permissions, audit logs, and environment-scoped egress and secrets.

Cursor 3.3 introduces a new PR review experience with inline threads, commit history, and file tree navigation. It adds parallel plan execution using async subagents and the ability to split changes into multiple PRs. Additional improvements include pinnable skill quick-actions, improved subagent model configuration, and a new /multitask command.

Cursor introduced multitasking with async subagents via /multitask command to parallelize requests, improved worktrees experience in the Agents Window for running isolated tasks across branches, and multi-root workspace support allowing a single agent session to target multiple folders for cross-repo changes.

This release introduces multitasking via async subagents using the /multitask command, allowing parallel processing of requests. It also adds improved worktrees support in the Agents Window for running isolated tasks across different branches, and introduces multi-root workspaces enabling a single agent session to make cross-repo changes spanning frontend, backend, and shared libraries.

Cursor 3.1 introduces interactive Canvases that allow the AI to respond with visualizations including dashboards and custom interfaces. The Agents Window received a tiled layout for running multiple agents in parallel, and voice input was upgraded with batch STT for higher-quality speech-to-text transcription.

Bugbot can now learn from feedback on pull requests and turn those signals into learned rules that improve future reviews, achieving a 78% resolution rate. MCP server support was added for additional context during code reviews on Teams and Enterprise plans. Additional improvements include a 'Fix All' action, redesigned settings, and improved Bugbot Autofix reliability.

Cursor added 30+ new plugins from partners like Atlassian, Datadog, GitLab, and Hugging Face, enabling cloud agents to read, write, and take actions across more of the development stack. Previous updates also introduced Automations for always-on agents triggered by schedules or events, JetBrains IDE support via Agent Client Protocol, MCP Apps with interactive UIs, and Bugbot Autofix for automatic PR fixes.

Cursor added 30+ new plugins from partners like Atlassian, Datadog, GitLab, and Hugging Face, enabling cloud agents to read, write, and take actions across more of the user's stack. Previous updates also introduced Automations for always-on agents triggered by schedules or events, JetBrains IDE support via Agent Client Protocol, MCP Apps with interactive UIs, and Bugbot Autofix for automatic PR fixes.

Composer 2 is now available in Cursor, offering frontier-level coding performance with strong results on challenging coding tasks. It introduces a two-tier pricing model: Standard at $0.50/M input and $2.50/M output tokens, and Fast (default) at $1.50/M input and $7.50/M output tokens. Additionally, self-hosted cloud agents are now supported, keeping code and tool execution entirely within the user's own network infrastructure.

Agent mode introduced. Cursor can autonomously run terminal commands, edit files and fix errors.

Claude integration added. Multi-file editing with Composer. Codebase-wide context understanding.

Initial launch as AI-first VS Code fork. Basic code completion and chat features.