Write one line.Claude sequencesthe rest.

Describe an app in one sentence. Claude Opus 5.5 edits the code with seven precise tools, a sealed sandbox verifies the build, and it goes live at its own address on writesequa.xyz.

How it works
AI app builder on Robinhood Chain

Prompt, live preview, deploy and token launch in one flow. Every app runs in its own container; every token is signed in your own wallet.

  • Claude Opus 5.5
  • Agentic edits
  • Verified builds
  • Its own address
  • A token on Pons V2

SEQUA reads your sentence like a sample. Claude isolates the files that matter, edits them with surgical tools, assays the build in a sealed container, repairs what fails, and expresses the result at its own address.

20
free AI runs a day
5
apps on the free plan

The Sequence. You write the first step.

Sample · 1 of 6
Say what you want

One sentence is enough to start. Add detail whenever you like.

Inside the agent. It reads only what matters.

Claude never receives your whole project. Like a sequencer reading one region of a genome, it gets the map, the conversation and the few files your request touches, then pulls in anything else on demand.

Seen as a path and size Read in full

What each request carries

  1. 1
    Project metadata

    Framework and settings, so every edit fits the stack.

  2. 2
    The file tree

    Every path with its size. No contents yet.

  3. 3
    The conversation

    A rolling summary of older turns plus the latest messages.

  4. 4
    The relevant files

    Only the files ranked closest to your request, within a size budget.

The tools it can use, and nothing else

  • list_filesLists every file in the project with its size.
  • read_fileReads one file in full, when the agent needs more than it was given.
  • edit_fileReplaces one exact snippet. The rest of the file is untouched.
  • write_fileCreates a file, or rewrites one almost entirely.
  • rename_fileMoves or renames a file.
  • delete_fileRemoves a file.
  • run_buildType-checks and bundles in the sandbox, and hands any errors back to the agent.

Spec sheet

Read live from this server's configuration
Model
Claude Opus 5.5
Writes and edits your code
Reasoning effort
High
How hard it thinks per turn
Steps per request
Up to 40
Tool calls in one run
Repair attempts
Up to 3
Automatic fixes when a build fails
Output per turn
64k tokens
Room for whole files at once
Prompt length
8,000 characters
Per message
Sandbox
512 MB · 1 CPU
Per project container
Build timeout
180 s
Before a build is stopped

Protocols

Built for ideas, not toolchains. What runs between your sentence and a live URL.

  • Talk, don't type code

    Ask for changes in plain words. Claude edits only what the request needs and leaves your design alone.

  • Watch it update

    A live preview refreshes after every verified build.

  • Ship in one click

    Redeploys swap atomically. A failed build never takes the live version down.

  • Your own address

    Every project lives at {name}.writesequa.xyz with a certificate issued on demand.

  • Isolated by default

    Each project builds and runs in its own container with CPU and memory limits.

Every run happens behind four walls.

Containment

  • Sealed builds

    No root, no host files. Each project gets its own container.

  • Secrets stay server-side

    Platform keys never reach generated apps or the browser.

  • You sign everything

    Your wallet, your signature. Nothing is signed for you.

  • Checked on-chain

    Launch results are read from the real receipt and verified.

Synthesis

Give your app its own token in one signature, on Pons V2, Robinhood Chain. Terms are read from the factory again in your browser right before you sign.

Live from chainPons V2 docs
$SEQUA launch termsConfig #0 of 1
Launch fee
0.0005 ETH
Creator tax
1% to 4%
Curve fee
1%
Supply
1B
Graduation
4.2 ETH
Chain id
4663
Factory 0x7eD598…01EC7e Read Sat, 31 Oct 2026 19:39:19 UTC
  1. Call 1
    canLaunch(wallet)

    Eligibility is checked before the form appears.

  2. Call 2
    previewLaunchEconomics(id, ETH)

    Pinned in your params: 0xa9fc75…8b1ca7

  3. Call 3
    launchToken(params) + fee

    Simulated first, then signed in your wallet.

  4. Call 4
    TokenLaunched(token, curve)

    Read from the receipt and verified server-side.

Reagents

Start free with 20 AI runs a day and up to 5 apps. Past that, credits are paid in $SEQUA at Claude's own token price plus 5%, nothing hidden.

Every account

AI runs, apps, and hosting on your own subdomain with HTTPS.

$0
$1 top-up
100 credits
$5 top-up
500 credits
$10 top-up
1,000 credits
$25 top-up
2,500 credits

Any amount works. You pay the dollar value in $SEQUA at the live on-chain price (now 1 $SEQUA ≈ $0.00000448).

Personal API keys

Create apps, run Claude and deploy from scripts, bots and other tools with an aestro_sk_ key. API requests always spend credits.

Get credits

Lab notes

Good questions, short answers.

  • Not at all. Describe what you want and Claude makes the change. You can peek at the code in a read-only view if you're curious.

  • Single-page apps built with React, TypeScript and Tailwind: landing pages, portfolios, dashboards, small games and interactive tools.

  • Each deployment runs in its own container and is served at {your-project}.writesequa.xyz with automatic HTTPS.

  • You do. SEQUA prepares the transaction from the current Pons V2 configuration, and your own wallet signs it. We never touch keys or seed phrases.

  • No. They're linked but independent. Editing your site later never touches the token's on-chain data.

Write one line