kash

Guide to getting the most out of coding agents (from the Cursor team): - plan before coding. when agent runs don't match what you wanted, revert changes and go back to the plan - let the agent find context, you don’t need to manually tag every file - start new conversations when switching tasks or if the agent seems confused - @ past chats instead of copy pasting whole conversations - rules = static context that applies to every conversation. start simple. add rules only when you notice the agent making the same mistake repeatedly. - skills = workflows agents can invoke when needed. you can set up hooks that keep agents grinding until all tests pass (see full guide below) - agents can process images directly from prompts (screenshots, design files, image paths). you can scribble on a piece of paper then share with Cursor - review review review - review during generation with the diff view - after the agent finishes, click review → find issues to run a dedicated review pass - use Bugbot for automated review on your PRs - for significant changes, ask the agent to generate architecture diagrams - run multiple agents in parallel, Cursor automatically creates and manages git worktrees for parallel agents. you can also run the same prompt across multiple models simultaneously - delegate tasks you’d add to a todo list to cloud agents (trigger agents from Slack with “@ Cursor”) - use debug mode when standard agent interactions struggle with a bug. provide detailed context about how to reproduce the issue. the more specific you are, the more useful instrumentation the agent adds. - devs who get the most from agents share a few traits: write specific prompts, iterate on their setup, review carefully, give agent verifiable goals, treat agents as capable collaborators

Great first week on Cursor CLI with @luist188 We shipped a lot Rough edges sanded down And here's a thread with a few new features we added Lots to still do, feedback always welcome

@cursor_ai @JonasBadalic Sentry 🤝 Cursor

Cursor Automations now support event-based triggers for Sentry. Set up agents that automatically respond to new issues, investigate root causes, open PRs with fixes, and post summaries to Slack.

Open up Cursor 3 and ⌘+D to start multi-tasking

We’re introducing Cursor 3. It is simpler, more powerful, and built for a world where all code is written by agents, while keeping the depth of a development environment.

join us tomorrow at 10am PT for a look at how we build always-on agents internally at Cursor!

it's a great model and it's only getting better

@potetotes @cursor_ai @reactjs @Meta welcome!!

This week I joined @cursor_ai! It's been incredible watching this small but mighty team and I'm excited to be a part of ushering in the third era of AI software development. Very grateful to the @reactjs team and @meta for the opportunity to have worked on React for the past 6 years. Shipping React Compiler at React Conf is going to be one of the major highlights of my career. I will hopefully continue to still be able to contribute to React in my own time.

We're increasing capacity for Composer 2, 2x more usage all weekend! Try it out!

Composer 2 is now available in Cursor.

Over 30 new plugins are now available on the Cursor Marketplace.

Cursor can now build agents that run automatically. Find a few examples to get started on cursor.com/marketplace

You can now use Cursor with 30+ ACP clients, including OpenClaw 🦞 This means complete access to Composer 1.5, codebase indexing and semantic search, and more! Here's an example with avante.nvim

you can now create and share plugins with your team with Cursor!

composer 1.5 is a good model, and now you're getting a lot more of it! sync work → composer 1.5, opus 4.6 async work → 5.3 codex

Will be doing an AMA tomorrow at noon PT! Will try to answer as much as I can about Cursor, models, AI coding, etc.

We trained the best coding model in the world under 1T parameters. Excited for people to try it!

Semantic search significantly improves coding agent performance. For very large codebases, Cursor's indexing process is now several orders of magnitude faster. cursor.com/blog/secure-co…

Cursor can now generate images using Nano Banana! Try it out in 2.4, this has been a lot of fun to use

Subagents are available in Cursor 2.4. Your agent can now spawn workers to tackle parts of a task in parallel. Each one gets its own context and can be configured with custom prompts, tools, and models.