# Cursor 1.0 turned code review and background execution into one editor release Cursor’s June 4, 2025 1.0 release combined BugBot review, broader Background Agent access, project memories and one-click MCP setup. Canonical: https://rangoon.ai/insights/archive-cursor-1-0-june-2025/ Author: Rangoon Editorial (https://rangoon.ai/insights/editorial/) Published: 2026-10-23 Updated: 2026-10-23 Event date: 2025-06-04 Tags: [Cursor](https://rangoon.ai/insights/tags/cursor/), [coding agents](https://rangoon.ai/insights/tags/coding-agents/), [code review](https://rangoon.ai/insights/tags/code-review/), [MCP](https://rangoon.ai/insights/tags/mcp/) Source note: Historical reporting: Cursor published its 1.0 changelog on June 4, 2025. The story records the release scope; it does not claim present-day availability, pricing or reliability. Image: Agent review receipt. Original Rangoon explanatory diagram: concepts for editorial review, not measured results, an official legal determination, vendor interface or event photograph. ## Key takeaways - Review comments need the same ownership rules as human review. - Remote agent output should arrive as a diff plus a test receipt. - MCP installation convenience does not replace connector review. ## A 1.0 release concentrated several workflow surfaces [source 1](https://www.cursor.com/en/changelog/1-0) [source 2](https://cursor.com/blog/june-2025-pricing) Cursor’s June 4 changelog called the release Cursor 1.0 and listed BugBot, general availability of Background Agent, memories, Jupyter agent editing and one-click MCP setup. Its June pricing follow-up is a second dated record that explains a move from request counting toward compute-based usage. Together they show a product widening from inline assistance to asynchronous work, review and external-tool connection. They do not establish that every plan, model or integration had identical access. ## The important boundary was between proposed code and accepted code [source 1](https://www.cursor.com/en/changelog/1-0) [source 2](https://cursor.com/blog/june-2025-pricing) BugBot’s release description says it comments on pull requests, while Background Agent works remotely and MCP setup can add authenticated servers. Those features place three different artifacts beside one another: a review finding, an agent-authored change and a tool connection. A delivery team should keep them separable. The reviewer needs the precise diff and reproduction path; the tool owner needs the server identity, grants and data boundary; the merge owner needs a human decision recorded outside the agent transcript. ## A useful receipt names the task, revision and test result [source 1](https://www.cursor.com/en/changelog/1-0) [source 2](https://cursor.com/blog/june-2025-pricing) For a background task, preserve the commit base, exact prompt, files changed, commands run, exit codes and unresolved questions. Treat a review comment as a finding until someone verifies it against the repository, rather than as evidence of a defect. For an MCP installation, capture the server URL, authorization mode and permitted methods before a contributor starts work. These records make a later rollback or incident investigation possible without relying on remembered chat context. ## Analysis: speed should make review narrower, not optional [source 1](https://www.cursor.com/en/changelog/1-0) [source 2](https://cursor.com/blog/june-2025-pricing) Editorial analysis: Cursor 1.0’s combination of remote work and review makes a narrow task boundary more valuable. Teams can gain iteration speed by assigning a bounded change, but should require the agent to stop at tests, generated files or configuration it cannot justify. A two-person review can focus on behavior and permissions when the agent receipt already identifies its command history. That is a local operating rule, not a claim about Cursor’s defaults. Cursor 1.0 made several distinct mechanisms visible at once. BugBot left a GitHub pull-request comment and offered a “Fix in Cursor” link with a prefilled editor prompt. Background Agent became available through the cloud icon or Cmd/Ctrl+E for users without privacy mode, while notebook editing could create and edit multiple Jupyter cells and initially supported Sonnet models. Memories were a beta, stored per project and per individual, and managed through Settings; one-click MCP setup was a separate connector installation path. These details create different records: a review comment, remote task, notebook cell edit, saved project fact and external server connection. Keep their owners separate. A BugBot finding needs a reproduction; an MCP connection needs its server identity and methods; a remembered fact needs a current repository check. The June pricing follow-up is usage context, not evidence that all product surfaces had identical access. ## Sources 1. [Cursor changelog: 1.0](https://www.cursor.com/en/changelog/1-0) 2. [Cursor: Clarifying our pricing](https://cursor.com/blog/june-2025-pricing) ## Related reading - [Cursor 2.0 framed parallel coding agents as a worktree coordination problem](https://rangoon.ai/insights/archive-cursor-2-0-october-2025/) - [OpenAI introduced SWE-Lancer to test coding agents against freelance software work](https://rangoon.ai/insights/archive-openai-swe-lancer-freelance-software-benchmark-february-2025/) - [GLM-4.7 made preserved thinking a named control for multi-turn coding agents](https://rangoon.ai/insights/archive-zai-glm-4-7-december-2025/) ## More information - [Documentation index](https://rangoon.ai/llms.txt) - [AI access and policies](https://rangoon.ai/ai/) - [Source repository](https://github.com/hypler-dev/rangoon)