Where infr is going

Let an AI assistant read and change your diagram, and run the same tests you run.

• Add, connect and edit components
• Run a failure test and read the findings
• Ask why something is a single point of failure

Run infr on your own network, with your diagrams on your own storage.

• One container, no outside calls
• Your sign-in, your storage

Turn a diagram into Terraform, Pulumi or Kubernetes manifests, and read existing state back into a diagram.

• Export what the catalog can describe
• Import state into a diagram

A monthly cost worked out from the plans and instance types in your diagram, and a side-by-side for two designs.

• Real prices, each with its source
• What-if between designs

More ways to test a design and better evidence behind every number the analysis gives.

• Correlated failures and maintenance windows
• Show how sure each figure is

Generate the checks that watch the design you drew, and compare what they see with what the analysis predicted.

• Probes for each link and service
• Reality against prediction

Snapshots of a diagram, and a visual diff between any two, including how the analysis changed.

• Restore any snapshot
• See what a change did to availability

A read-only connection that draws what you actually run on AWS, GCP, Hetzner and others.

• Read-only access only
• Spot drift from the diagram you meant

Ramp traffic, add latency, break things over time and watch where the design gives way first.

• Traffic ramps
• Failures that arrive in sequence

Name a failure you care about, keep it with the diagram and re-run it after every change.

• Power cuts, lost zones, lost links
• See which ones started failing

Turn a failure the analysis found into a recovery checklist for that exact design.

• Steps in the order things come back
• Export as text

Recovery time and data loss worked out from your backup and replication links, per component.

• What a restore depends on
• Where the longest wait is

Run the analysis on a diagram file in your own pipeline and fail the build when a new single point of failure appears.

• Works on .infr.json files
• Plain text and JSON output

Trace every route from the internet to your data and see which one control closes the most of them.

• Paths through links and zones
• What a firewall rule would cut

Place zones on a map and have latency between them worked out from real distance, then see what it does to a request.

• Cross-region links priced in milliseconds
• Pick regions by who your users are

Power draw from catalog figures, turned into energy over a month and an estimate of emissions.

• Built on vendor-rated figures
• Compare two designs

Shared libraries, roles, and comment threads pinned to components.

• Review a design in place
• Roles per diagram

Flag open ports, unprotected data and single-region storage against common baselines.

• Every check names the rule behind it

Put a live diagram or an analysis result in your docs and READMEs, and read it over an API.

• Embed a diagram
• Fetch findings as JSON

See what others have suggested, vote on it, and follow each idea from sent to planned to shipped.

• Vote instead of repeating a request
• A status on every suggestion

Send the diagram you are looking at with a problem report, so the person reading can open exactly what you saw.

• Optional, and you see what is sent
• Size-limited

Leave an email on a suggestion and hear once, when it lands in the changelog.

• One message, no newsletter
• Unsubscribe from the message itself

Roles between owner and member, such as admins who can invite and remove people while billing stays with the owner.

• Admins manage people, not billing
• Readers who can open but not change

In a live session, choose person by person who can only watch and who can edit, instead of one setting for everyone.

• Change someone's access mid-session
• Guests watch, teammates edit