Working with Syntera's Sandbox Environment

The sandbox is a containerized environment with pre-installed tools and frameworks. When Syntera creates your project, it:

• • Provisions a new Daytona sandbox from a snapshot
• • Clones your repository (or initializes a new one)
• • Installs dependencies based on framework detection
• • Generates a codebase tree for navigation
• • Sets up environment variables and database credentials

Sandbox Layout
/home/daytona/
  ├─ your-repo-name/     # Cloned repository
  │   ├─ src/
  │   ├─ package.json
  │   └─ ...
  ├─ .env                # Auto-configured env vars
  └─ [pre-installed tools]

Syntera uses framework-specific snapshots to minimize setup time. A Next.js project gets a snapshot with Node.js, npm, and common React dependencies pre-cached.

The sandbox provides full filesystem access through Syntera's tool layer. Every file operation goes through a typed backend protocol:

• • Read files with line-number offsets for precise navigation
• • Write new files with automatic Prettier and ESLint formatting
• • Edit existing files with pattern-based string replacement
• • Search files with glob patterns and content grep
• • List directories with metadata (size, type, permissions)

// What happens when Syntera writes a file
async write(filePath: string, content: string) {
  // 1. Write the file content
  // 2. Run Prettier to format
  // 3. Run ESLint --fix for auto-corrections
  // 4. Return quality report (errors/warnings)
}

This means every file Syntera produces is consistently formatted and lint-clean, regardless of which subagent generated it.

Syntera can run shell commands in the sandbox for tasks like installing packages, running builds, and starting dev servers. All commands pass through a security filter first.

• • Package installation (yarn install, pip install)
• • Build and compile (yarn build, tsc)
• • Dev server startup (yarn dev on configurable port)
• • Test execution (yarn test, pytest)
• • Git operations (commit, push, branch)

Dangerous commands are blocked automatically - things like rm -rf /, chmod 777, curl piped to bash, and system shutdown commands will never execute.

Each sandbox includes a PostgreSQL instance. Syntera manages credentials automatically and can create schemas, seed data, and run migrations as part of the build process.

// Auto-generated database configuration
const dbConfig = {
  host: process.env.DB_HOST,
  port: parseInt(process.env.DB_PORT || '5432'),
  database: process.env.DB_NAME,
  user: process.env.DB_USER,
  password: process.env.DB_PASSWORD,
};

Database credentials are cached per project to avoid redundant API calls and reduce setup latency on follow-up requests.

Once your application is running in the sandbox, Syntera exposes a live preview URL so you can see your project in a browser without deploying it anywhere.

• • Automatic dev server detection and startup
• • Configurable port (default: 3000)
• • Preview URL accessible from your browser
• • Hot reload support for real-time changes
• • Framework-specific dev commands auto-detected

The sandbox is what makes Syntera safe and fast. Your code is written, tested, and previewed in an isolated cloud environment before it reaches your repository. Understanding this gives you confidence that Syntera's changes won't affect anything until you approve and merge the pull request.