Please do not open a public GitHub issue for security vulnerabilities.

Report security issues privately via one of these channels:

• GitHub Security Advisories — Report a vulnerability (preferred)
• Telegram — @zyvadev

Include as much detail as possible:

• Description of the vulnerability and its potential impact
• Steps to reproduce
• Affected version(s)
• Any suggested fix or mitigation

You will receive an acknowledgment within 48 hours and a resolution timeline within 7 days for critical issues.

ZYVA Cloud IDE is built with the following security properties:

• All AI inference runs on 0G Private Computer inside a Trusted Execution Environment (TEE). Prompts and responses are not logged by the inference provider.

• User code runs in isolated WebContainer (browser WASM) or on-demand E2B sandboxes — never on a shared server process.
• Every user's workspace is scoped to their own directory (/workspaces/{userId}/). Path traversal outside this boundary is rejected server-side.
• Terminal commands go through a policy layer: allow / approve / deny before any execution.

• Sessions use JWT HttpOnly cookies via NextAuth v5. Tokens are never exposed to client-side JavaScript.
• OAuth tokens (GitHub, Google) are stored server-side in PostgreSQL and never returned to the browser.
• Wallet sign-in uses EIP-191 personal_sign (SIWE) — no private keys are sent to the server.

• All secrets live in .env.local (gitignored) and are never shipped in client bundles or exposed over HTTP.
• The repository does not contain real API keys, database credentials, or private keys.

• Desktop builds are unsigned. Windows and macOS will show a first-launch warning. Only install builds from official releases or zyva.dev.
• TEE attestation is reported honestly — ZYVA does not claim verified remote attestation until a full quote flow is implemented.

In scope:

• Authentication bypass or session hijacking
• Cross-user workspace access (path traversal, insecure direct object reference)
• Remote code execution on the application server
• Credential or secret exposure in the codebase or builds
• XSS in the IDE or docs that could exfiltrate session tokens

Out of scope:

• Vulnerabilities in third-party services (0G, E2B, GitHub, Google)
• Issues in unsigned desktop builds that require physical access
• Social engineering or phishing
• Denial of service against the preview deployment

We follow coordinated disclosure. Once a fix is deployed we will publish a security advisory crediting the reporter (unless they prefer to remain anonymous).