I can relate to this.
Highly reccomend reducing your workflow down to 2-3 agents.
Use 1-2 as executors. And 1 as a personal tutor that explains architecture and data flows to you as you build.
Ask it questions. Go deep. Sketch stuff out on paper. Build. Learn. Repeat. Take it slower than you think. It’s okay to leave agents waiting.
As builders we have forgotten that what often drives us is our curiosity and desire to learn, and that is taken from us when we are trying to maximize agent output.
You can outsource your thinking to ai, but you cannot outsource your understanding.
We’re introducing the Cursor SDK so you can build agents with the same runtime, harness, and models that power Cursor.
Run agents from CI/CD pipelines, create automations for end-to-end workflows, or embed agents directly inside your products.
Thanks for the comment! - not quite putting Cursor inside Gmail - more like using Gmail/Chat as the intake surface for code work.
flow is:
-share an email/thread into Chat,
-the app classifies what kind of request it is
-a Cursor SDK agent can pick up the actual engineering task: bug investigaiton, doc update, draft a fix, turn thread into working pr, etc...
The interesting bit is moving from “someone emailed me context” to “an agent is working on it." and I'm sure it could helpful in email responses as well!
The biggest difference IMO is that Cursor SDK isn’t just “call an LLM with tools.” It’s exposing the same coding-agent runtime Cursor already uses: repo context, edit/search/terminal workflow, streaming status, model choice, and local/hosted execution.
For this demo, Gmail/Chat is just the intake/collab layer. Cursor SDK is the part that can actually go operate on a codebase like a Cursor agent would. (in short: the harness)
ty! I actually avoided the browser-extension path. The trick is using Gmail’s native “share in Chat” flow, then letting a Google Chat app pick up the context and route it through theCursor SDK.
Much less brittle than injecting UI into Gmail, and the result lives in a thread I can share with people on my team!
@christianjesusz probably not a browser extension tbh. the whole thing I like here is that Gmail already has a “add to Chat” button.
The user can push an email thread into a real workspace conversation and the agent can work from there. feels much cleaner than injecting UI into Gmail.
How it works:
→ Gmail “add to chat”
→ google chat opens and i tag my agent "@uma" (custom google app)
→ pulls in the email thread + user intent
→ Gemini Flash Lite routes the request
→ Cursor SDK hands it off
→ Cursor Cloud Agents do the work
→ Results get streamed back to google chat app
@pranaviyer27 I love nyc. Being in tech is awesome but I think I would get sick of it if it I lived in the sf bubble. Staying close to friends in finance, fashion, theatre, law etc… is something I’m grateful for everyday