Skip to content
braincell.sh

Build, verify, ship. Documented.

Reference for the braincell platform. The agent loop, the microVM contract, snapshots, Universal Key, deployments, and the surface area we expose for power users.

braincell/Docs · Sandboxes
v2.4
Quickstart
Agent loop
Sandboxes
Universal Key
API reference
Boot a fresh microVM
const sandbox = await
braincell.sandbox.create(
{ project: 'pomodoro', persist: true });

Quickstart

From signup to a deployed URL in under 10 minutes. The shortest path through the agent loop.

Start the walkthrough

The agent loop

Five phases the agent owns end to end: describe, design, build, verify, deploy. Read what each one does.

Read the loop

Universal Key

One key, every frontier model. Drop-in compatible with the OpenAI SDK, billed against your braincell credits.

Wire it in

Models

Opus 4.7, Sonnet 4.6, Haiku 4.5, plus GPT-5 and OpenRouter fallbacks. How to pick the right one per turn.

Compare models
By role

Find the page that matches your job.

The agent loop looks different depending on what you ship. Pick your seat at the table.

Product managers

Ship a prototype before the spec doc. Validate ideas with real users on Friday.

Designers

Push your taste straight to production. No handoff to engineering, no half-built mockups.

Engineers

Skip the boilerplate phase. Clone the project locally when you need to take over.

Founders

Stop paying for SaaS you barely use. Replace a stack of subscriptions with one tool you own.

The agent loop

Every braincell turn runs five phases: describe, design, build, verify, deploy. The agent owns each phase. You only step in to confirm the result. Five strikes per turn means typecheck, lint, and a real build all pass before any change reaches you.

braincell/Agent · Turn 14
Running
Describeprompt parsed
Designschema + routes
Buildcode generated
4Verifytypecheck + lint
5Deployvercel · prod

Universal Key

One key proxies to Anthropic, OpenAI, Google AI, and OpenRouter. Billed against your braincell credits, no separate provider accounts required. Drop-in compatible with the OpenAI SDK, so the migration is a one-line base-URL swap. Use it inside braincell apps or paste it into your own stack.

braincell/Universal Key
OpenAI compatible
Your key
bc_live_8aJk··········42pQCopy
// drop-in for the OpenAI SDK
const client = new OpenAI({
baseURL: "https://braincell.sh/v1",
apiKey: process.env.BRAINCELL_KEY,
});
Anthropic
OpenAI
Google
OpenRouter

Snapshots and rollback

The agent captures the full project state on every assistant turn. Rewind any decision with a single click. Snapshots are diffable, named, and free on every paid plan. Use them as A/B variants, undo a regression, or fork a working app into two directions.

braincell/Audit log
Streaming
mira@Enabled SSO · SAML2m ago
systemAudit log → S3 export ran14m ago
agentDeployed booking-app to prod1h ago
priya@Added seat · ops team3h ago
Runtime

Sandboxes (Vercel microVM)

braincell does not render previews in a browser stub. Every project boots a real Vercel microVM. Your install script runs, your env vars resolve, your routes hit real Node. If something fails in production, it fails in the sandbox first.

Models

Supported models

  • · Claude Opus 4.7: default, deep multi-step reasoning
  • · Claude Sonnet 4.6: balanced quality and cost
  • · Claude Haiku 4.5: fast tool-calls, low credits
  • · GPT-5 and OpenRouter fallbacks for niche tasks
Tooling

Command line

A CLI ships alongside the workspace: clone a project locally, run the sandbox commands, push back to braincell. Targeting public beta in 2026.

Common questions.

Reach out at support@braincell.sh for anything beyond this list.

Nothing to install. braincell runs in the browser as a hosted workspace at app.braincell.sh. A CLI is in private beta for teams that want to clone projects locally.

Stop reading. Start shipping.

The fastest way to learn braincell is to ship one app. Ten credits, no credit card.

10 free credits on signup. No credit card.