Launched this week
Grass
Gives your coding agent a dedicated VM that's ready 24/7
224 followers
Gives your coding agent a dedicated VM that's ready 24/7
224 followers
Grass gives your coding agent a dedicated VM that's always ready. No setup, no config, no burning your laptop. Point Claude Code or OpenCode at it and run. Monitor progress from your phone, steer mid-session, push changes, all without touching your machine. Every new account gets 10 hours free. No credit card needed.
Grass
Hey PH!
We built Grass because we noticed something weird. We message our teammates from anywhere, on the train, at lunch, from the couch. But the moment we want to talk to our coding agent, we need a laptop and a terminal open.
That felt broken. Coding agents aren't tools you operate anymore. They're colleagues you work with. Grass is how you stay in touch with them.
What Grass does today:
→Thread continuity across your phone and laptop. Start a task on Claude Code or OpenCode on your machine, pick up the same conversation on your phone. Same context, nothing lost.
→ Free pre-configured VM. We sponsor 10 hours of Daytona compute. No credit card, no DevOps setup. Your agent has somewhere to work in under a minute.
→Bring your own API key. We never see it, never store it. Your keys stay on your device.
→Agent agnostic. Works with Claude Code and OpenCode today. More agents coming.
On security: Grass runs a BYOK architecture. Your API key never touches our servers. We're building for developers who care about this, because we're developers who care about this.
Why we made it free to start: Every developer running coding agents should be able to try cloud compute without a credit card form in the way. 10 hours is enough to have your first "my agent finished while I was away" moment. After that, you'll know if Grass is for you.
A bit about us: Raunaq previously built Elemential (enterprise infra, shipped to NSE) and has trained 3,000+ developers on AI agent infrastructure. Anil built the core architecture, real time sync, SSE-based event buffering for mobile reliability, and Daytona VM orchestration. I lead design and creative, everything around how Grass shows up in the world.
We'd love your feedback, especially on what agents you want us to support next and what your ideal mobile coding workflow looks like. We're in the comments all day.
@sunnyjoshi Hi. Does the VM stay ready 24/7 even if I close the app or switch devices? Any downtime?
Grass
@julia_zakharova2
That's the whole point of Grass. The VM doesn't care if your phone is in your pocket or charging on your nightstand. Your agent keeps shipping.
However, If nothing's happening for 15+ minutes, like no agents running and the VM is idle, it'll sleep automatically to save resources. Wakes back up within 5 seconds though.
Product Hunt
Grass
@curiouskitty Great question. We gate execution at the tool level. Anything that mutates state (file writes, shell commands, installs, etc.) triggers an approval.
When that happens, we show the exact command/diff, along with relevant context (what the agent is trying to do and why). We’re also starting to layer in simple risk signals (e.g. destructive commands, wide file changes) to help users make quicker decisions.
To avoid stalls, we support session persistence + async approvals. So you can approve from your phone, or set scoped auto-approvals for low-risk actions. We’re also exploring fallback behaviors (like safe retries or partial execution) when a user is away.
Very cool, does this mean I can finally touch grass? 😄
Jokes aside, how are you thinking about session continuity and memory when there’s a disconnect or the user switches devices?
Grass
@jon_dalgir Haha, that’s the goal 😄
Right now, your local machine (tethered) and the cloud VM are treated as separate environments, so sessions don’t automatically carry over between them.
We’re working on ‘Grass Sync’ next, which will persist both the codebase snapshot and agent threads, so you can switch devices or reconnect without losing context. That’ll also open up sharing and collaboration across teammates.
@anildukkipatty That makes a lot of sense, appreciate the honest answer. “Grass Sync” is a great name too 😄 Preserving both the code snapshot and the agent thread context feels like the right unlock, especially for switching devices and teammate collaboration.
GAIA
Pretty cool launch guys! How does this differ than claude code remote control? Is it the fact that it's part of a hosted VM instead?
Grass
@aryanranderiya Thanks! Yeah, it's part of a hosted VM. Grass is a persistent hosted workspace. Your agents run in a sandbox that stays alive, so you can jump in from your phone anytime (no laptop needed).
It’s multi-agent Claude Code, OpenCode today. Adding Codex, pi next. Which ones would you want to try?
Macky
Interesting, how does it connect to the VM from iPhone? does it ssh or just a message relay?
Grass
@sayuj_suresh It’s a message relay. The Grass server runs the session and handles communication between your phone and the VM no direct SSH from the iPhone.
I love the idea of having a dedicated VM to do work and maintain context. This sounds like having the claude code app, but running 24x7. What kind of VM is being generated and persisted in the background (i.e. EC2s, docker container etc?)
Best of luck on your launch day, I'm running mine as well! I'm using claude and will give this a try.
Grass
@gparker97 Thanks, and good luck with your launch too! will check it out.
Under the hood, each workspace runs in an isolated Docker container. We’re currently using Daytona to spin up and manage these sandboxes, which lets us keep them lightweight but still persistent enough for long-running agent sessions.
Okan
Running a local coding agent usually ends up chewing through all my RAM and crashing my active dev environment. Spinning up a persistent VM specifically for this is a brilliant way to handle long-running background refactors. I would love to know if you offer pre-configured images for different agent frameworks out of the box.
Grass
@y_taka Local agents can get overwhelmed, especially for long-running tasks.
Right now, we offer pre-configured environments for Claude Code and OpenCode out of the box, with Codex and Pi coming next.
We’re also thinking about this more as ‘agent-ready workspaces’ rather than just base images. So the goal is to have setups that are tuned for specific workflows/frameworks over time.
Curious, what coding harness or setup are you using today?