Skip to content
Security

Built for teams, secured for enterprise

Every control a security team checks for, built in from day one — not bolted on at renewal time.

  • OAuth 2.0 PKCE
  • AES-256-GCM at rest
  • OpenAPI 3 spec
  • Server-side RBAC
  • SOC 2 Type II (in progress)

Data isolation and tenancy

Every organization is isolated at the database level, and every project inside it has its own data scope and access boundary.

  • Cross-project data never leaks, even for admins switching between teams in the same organization.
  • Every API call is scoped on the server: even a super-admin on the wrong subdomain only sees that tenant.
  • No client-side trust: permissions are checked on every request.

Encryption and your keys

API keys and bring-your-own-key credentials are encrypted at rest with AES-256-GCM and only decrypted per request.

  • On the Enterprise plan, bring your own Anthropic key: AI spend hits your own account and CloudByte never proxies billing.
  • Managed or self-managed key tiers, depending on how your team prefers to run billing.
  • Traffic between the agent, the dashboard and the API is encrypted in transit.

Identity and access

Google SSO with OAuth 2.0 PKCE and domain restriction per organization, plus passwordless magic links whose one-time tokens expire in 5 minutes.

Roles and access
RoleAccess
Super AdminFull control across every organization on the account.
Org AdminFull control within one organization: users, roles, integrations and settings.
ReviewerReview access to dashboards, skills and alerts for the projects they are assigned.
StandardDay-to-day access to their own activity and the team views their admins allow.

What the agent collects, and what it never does

A 3 KB agent reads the local session data of Claude Code, Cursor and GitHub Copilot without modifying it and captures git commits from the repositories you point it at.

Collected

  • Prompts and tool observations
  • Session details and summaries
  • Commit metadata and diff stats
  • Commit diffs for summaries (kept 5 days)
  • Token usage per model
  • Agent heartbeats and skills usage

Never collected

  • Source code files
  • Keystrokes
  • Environment variables
  • Screen recordings
  • Browser activity

Sensitive prompt alerts

Prompts containing secrets, credentials or PII are flagged before they leave the developer machine.

  • Blocked or flagged by severity, with real-time alerts to admins.
  • Every alert is kept in the audit trail for compliance review.
  • Catches what developers paste by accident: keys, tokens, passwords and personal data.

Audit and compliance

Every AI action is logged with user, organization, prompt and token spend, so you can answer who sent what to which model.

  • CLAUDE.md compliance tracking per developer and per project.
  • Every dashboard feature is also a REST endpoint with an OpenAPI 3 spec, so audit data can flow into your own tools.
  • SOC 2 Type II is in progress.

Deployment options

Run on CloudByte cloud, or keep everything inside your own environment on the Enterprise plan.

  • CloudByte cloud

    Fully managed, with data isolated per organization and per project.

  • Self-hosted

    A single Next.js and Express service on PostgreSQL, containerised and Terraform-deployed. Enterprise plan.

  • Air-gapped

    No outbound access, with in-org model endpoints such as Bedrock or a private gateway. Enterprise plan.

Retention and your control

Data retention follows your plan: 7 days on Free, 90 days on Starter, unlimited on Team, and a custom policy on Enterprise. Commit diffs are always deleted after 5 days, while metadata and AI summaries stay for your history.

  • Admins decide who is monitored and who can see what.
  • Ask us to export or delete your organization data at any time.

Responsible disclosure

Found a vulnerability? Email hello@cloudbyte.ai with "Security" in the subject line. Please give us a reasonable chance to fix the issue before sharing it publicly, and do not access data that is not yours.

For questionnaires and procurement reviews, use the contact form and choose the security topic.

Bring AI coding inside your security program.

Walk through isolation, RBAC, prompt alerts and deployment options with our team, or start free and review the data captured yourself.

  • Free for up to 5 developers
  • 10-minute setup
  • Your Anthropic key, your spend (Enterprise)