01
Conversational, not configurational.
Describe what you want — import, modify, remove. The agent drafts the work; you review and approve. No YAML, no state files, no second language.
the intelligent ide for your cloud
Sketch your infrastructure.
Ship it.
SkySurf is an AI-native IDE for AWS. Talk to an agent, watch your infrastructure render on a paper canvas, and approve every change before it lands.
SkySurf · acme · canvas
live · 6 resources · acct staging
fig. 01 · the agent proposes an instance-type upgrade, renders the diff, and waits for approval
what it is
A drafting table for the cloud, run by an agent you can talk to.
02
Every change, on paper first.
Every mutation lands as a reviewable change set with a plain-English summary and a machine-readable diff. Nothing touches AWS until a human clicks approve.
03
Start fresh, or import what you already have.
Sketch a brand-new architecture on an empty canvas, or point SkySurf at an existing AWS account and pull in whatever you need — one service, one tier, or the whole estate. Import is a feature, not the point.
how it works
From request to rendered resource, in four unhurried steps.
- step 01connect
Drop in an AWS account.
Paste an access key (read-only is fine to start), give it an alias like prod or sandbox. Credentials are AES-GCM encrypted with a key the platform never sees in plaintext.
- step 02ask
Tell the agent what you want.
“Import every S3 bucket from prod.” “Show me EC2 instances not behind a load balancer.” The agent picks the right tools and drafts a change set.
- step 03approve
Review the change set.
Plain-English summary up top, diff underneath. Impact line tells you how many resources and how many are destructive. Approve, reject, or keep iterating.
change set · 4a9fbd62awaiting approvalImport 4 S3 bucket(s) from prodimport 4 resources · 0 destructive- import s3:bucket arn:aws:s3:::public-assets
- import s3:bucket arn:aws:s3:::billing-exports
- import s3:bucket arn:aws:s3:::tf-state-prod
- import s3:bucket arn:aws:s3:::ingest-raw
approvereject - step 04render
Watch it land on the canvas.
A background worker executes the approved change set against AWS, reconciles local state, and the new nodes ink-bleed onto the graph. Drift is flagged, not hidden.
how it's different
The lay of the land.
Every other tool solves part of the problem and leaves the rest on your team. SkySurf is the one your whole team can actually use.
cross-reference · fig. 02 · tooling comparison
scale 1 : 1| property | AWS Console clickops | Terraform / Pulumi infra as code | Railway / Vercel opinionated paas | SkySurf conversational ide you are here |
|---|---|---|---|---|
| Who can maintain it | Whoever has the root keys | Your platform team, alone | Any app developer | Anyone on the team |
| Onboarding a new engineer | Tribal knowledge & a lot of clicks | Weeks learning DSL, modules, state | Fast — inside their walls only | Open the canvas, see everything |
| Picture of your whole system | Two hundred browser tabs | Buried across .tf files | A flat list of services | Live visual canvas |
| Making a change safely | One wrong click breaks prod | PR + plan + hope | Preview deploys, web tier only | Diff, impact, approve — every time |
| Working with infra you already have | It's there, just hard to see | Manual imports, weeks of work | Migrate to their stack, or skip it | Point at your account, instantly mapped |
| Many accounts, one mental model | Switch roles, lose your place | A workspace per account | — | Every account on one canvas |
| Knowing what changed and why | Dig through CloudTrail | Read the git log | Per-service deploy logs | Every change-set, narrated |
| Talking to it like a colleague | Not really | No | Bolted on | Built for it |
drawn: april 2026 · subject to revisionsheet 02 / 02
put your cloud on paper