 went viral this week. It runs on your own machine and you can connect it to WhatsApp, Telegram, Discord, or whatever chat app you already use. The project 

, and sparked a wave of community excitement. 

Of course, giving AI full access to your machine is not the greatest idea, so several folks have already created Railway templates to make it possible to run it remotely instead. We also put together 

 showing how to deploy your own and configure it (we went with 

Now for the stuff we built. It's a big one this week. Priority Boarding gets Postgres metrics and network flow visualization (pretty orbs and purple pipes on your canvas). Login with Railway brings proper OAuth to the platform. Volumes can resize without restarts. And templates can now pull from private Docker registries.

View database metrics for your Postgres database

You can now view detailed metrics for your Postgres databases directly in the Railway dashboard. See connection counts (active, idle, idle in transaction), cache hit ratios, query statistics with timing breakdowns, and table-level information including row counts, data sizes, and index usage. 

The Query Statistics panel surfaces your most expensive queries with call counts, rows returned, and timing percentiles. The Tables view shows you exactly where your data lives and how it's being accessed.  If you're running Postgres on Railway, this gives you the observability you need without reaching for external monitoring tools. Let us know what you think on 

Your canvas now shows live network traffic between services. Animated flows visualize the actual packets moving through your infrastructure, with the thickness indicating throughput.

Toggle between network flows and reference variables using the canvas controls in the top left.

Under the hood, this is layer 4 network observability. Each flow is aggregated by source IP, source port, destination IP, and destination port every 5 seconds. You can inspect latency, see which peers are connected, and find out why packets were dropped.

Head to a service's logs and you'll find a new Network Flow Logs tab where you can dig into the raw data: protocol, source and destination addresses, connected peers, traffic volume, latency, and status.

Platforms can now authenticate users with their Railway account

We want to make it easier to extend Railway, both for developers building tools and for platforms building integrations.

Until now, platforms integrating with Railway had to ask users to manually provide an API key. That meant users had to navigate to their account settings, create a token with the right scope, copy it, paste it into the integrating platform, and hope they didn't mess up the permissions. If the token got invalidated? Back to step one.

Login with Railway changes this. It's a proper OAuth 2.0 implementation with OpenID Connect. Users click a button, approve the permissions they want to grant, and they're done. The platform gets secure, scoped access without ever touching raw API keys.

Create an OAuth app in your workspace settings under 

Redirect users to the authorization endpoint

Exchange the authorization code for tokens

Use the access token to make API requests on behalf of the user

The system supports both web apps (confidential clients with client secrets) and native apps (public clients using PKCE). Refresh tokens are available for long-lived access, and users can select exactly which workspaces or projects to share.

For developers already building Railway integrations, this is the path forward. For users, it means fewer API keys floating around and more granular control over what third-party tools can access.

Resize volumes without restarting your service

Previously, resizing a volume meant restarting your service. For most workloads that's fine, but for long-running processes or services where uptime matters, it was a pain point.

Live Volume Resizing attempts to resize your volume without any restart. If the live resize fails (which can happen depending on the underlying filesystem state), a restart will be required as a fallback. Either way, you're in control.

To try it out, head to your volume settings and look for the 

 option. You'll be asked to type a confirmation phrase before proceeding. This is a one-way operation, after all.

Templates can deploy from private Docker images

Templates can now deploy services from private Docker images

Template authors, this one's for you. You can now include private Docker images in your Railway templates by securely sharing registry credentials.

When someone deploys your template, the services built from private images will work seamlessly. The user never sees or accesses the underlying credentials. For added security, SSH access is blocked for these services, keeping your credentials and container internals protected.

This opens up new possibilities for distributing proprietary software, internal tools, or commercial applications while benefiting from 

We shipped an improvement for deployment messages with 

 to set a deployment message that shows up in the UI the same way a commit message would

We improved the Service Settings page with a new "Scale" section that brings together regions, replicas, and replica limits all in one place

https://res.cloudinary.com/railway/image/upload/v1769815283/2026-30-1_nichoo.png

DB metrics, network flows, login with Railway, zero-downtime volume resizing

Changelog #0275