Security is built in, not bolted on

CloudCode is designed with security at every layer — from the database to the browser sandbox.

TLS 1.3Encryption in transit
AES-256Encryption at rest
MFAMulti-factor auth
Zero-accessCode never mined
SOC 2In progress
GDPRCompliant

Infrastructure security

Cloud-hosted on hardened infrastructure

CloudCode runs on Vercel, and its database is Neon Postgres. Both platforms provide DDoS protection and network-level firewalls, and every database query runs under row-level security for the signed-in user.

Encrypted in transit and at rest

Every request between your browser and CloudCode is encrypted with TLS 1.3. All stored data — project files, auth tokens, environment variables — is encrypted at rest with AES-256.

Protected environment variables

Environment variables are stored separately from your project files, behind row-level security so only you can read them. Values you mark Secret are masked in the UI and replaced with placeholders in the .env.example CloudCode generates for your repo.

Your code runs in isolation

The dev preview bundles and runs your app inside a sandboxed iframe in your own browser, and the terminal is a lightweight shell over your project files rather than a shared server. Commands the agent runs, and Server previews, run in Vercel Sandbox microVMs; the agent's sandbox gets none of your environment variables and can reach only package registries and GitHub. Deployments run in Vercel's isolated infrastructure.

Connected databases

A database you connect is reached over TLS only, never at a private or internal address, and its connection string is encrypted and never shown again or sent to the browser. The agent applies only migrations that add or change structure; anything that deletes data waits for you.

Authentication & access control

Multi-factor authentication (MFA)

CloudCode supports passkeys (Touch ID, Windows Hello, a phone or a security key) and TOTP two-factor authentication with an authenticator app (Google Authenticator, Authy, 1Password). Add them in Settings → Sessions & Security; accounts with a password confirm it to turn two-factor on or off. We strongly encourage enabling it, especially on accounts with production deployments.

Sign in with Google, Apple, or email

Sign in with email and password, or with Google or Apple through 1Profile using standard OAuth 2.0 / OpenID Connect flows. Sign-in only asks for your basic profile and email address. Your signed-in session is kept in a secure session cookie.

JWT-secured API

API calls are authenticated with short-lived ES256 JWTs minted from your signed-in session. Each token expires after 15 minutes and there are no refresh tokens: a new token comes only from a live session, so signing out stops new ones from being issued.

Row-level security

Our database enforces row-level security policies on every table. Even at the database layer, a user's data is inaccessible to other users — not just at the application level.

Data protection & privacy

Your code is never used for model training

We do not send your project files to any AI provider for training purposes. Project context is sent to the AI model only to answer the prompt you sent.

GitHub access, explained

Connecting GitHub is optional and separate from sign-in. CloudCode requests the repo and read:user OAuth scopes. GitHub's repo scope grants read/write access to all repositories your account can access (GitHub doesn't let OAuth apps limit it to selected repos). We never request org-admin, billing or account-management scopes, your token is stored server-side, and you can disconnect at any time.

Data residency

Project data is stored in the US, in Neon's US East (iad1) region. EU data residency isn't offered yet. Contact us if this is a hard requirement for your team.

No advertising trackers

We measure page views with Vercel Web Analytics and Viziera. There are no advertising trackers or pixels on any CloudCode page.

Internal access controls

Least-privilege access

CloudCode employees are granted the minimum access required for their role. Production database access is audited and requires approval via an access-request workflow.

MFA required for all staff

Every CloudCode team member uses hardware security keys or TOTP MFA on all internal systems. VPN access to production infrastructure requires a hardware token.

Audit logs

All production access is logged with timestamps, actor identity, and action details. Logs are tamper-evident and retained for 1 year.

Application security

Sandboxed preview execution

The Sandpack dev preview runs inside a sandboxed iframe with strict Content Security Policy headers. JavaScript running in your preview cannot access the parent CloudCode application.

CSRF and XSS protection

All API endpoints include CSRF tokens. User-generated content is sanitized before rendering. Our CSP prevents inline script injection.

Dependency auditing

We run automated dependency vulnerability scans on every build. Critical CVEs are patched within 24 hours; high severity within 7 days.

Regular penetration testing

We conduct third-party penetration tests at least annually and after major infrastructure changes.

Vulnerability disclosure

Responsible disclosure policy

We welcome security researchers who discover vulnerabilities in CloudCode. Please report issues to cloudcode@aw3.tech before public disclosure, giving us time to remediate.

Response SLA

We acknowledge all vulnerability reports within 24 hours. We aim to patch critical vulnerabilities within 48 hours and high severity within 7 days.

Safe harbor

We will not pursue legal action against researchers who follow our responsible disclosure policy. We ask that you do not access or modify other users' data during testing.

Found a vulnerability?

We take every security report seriously. Email us and we'll respond within 24 hours. Responsible reporters are credited in our security acknowledgements.

cloudcode@aw3.tech