Security
How we protect your work
Gendora handles requirements, deliverables and the API keys for your model providers. This page describes the safeguards in place today and the limits that still apply during Beta.
Workspace isolation
All business data belongs to an organization (workspace). Every query goes through a repository layer that enforces the organization condition, so one workspace can never read another’s projects, tasks or deliverables.
Requests for resources in a workspace you do not belong to return “not found”, so outsiders cannot even tell whether they exist. Within a workspace, owners, admins and members have different permissions, and membership changes are recorded in an audit log.
Encrypted provider keys
API keys for your model providers are encrypted at rest with AES-256-GCM using a server-side master key. They are decrypted only when a model call is made and are never returned to the browser: the interface shows a masked value.
Provider addresses are checked before every request. On the hosted service, addresses in private and local networks are refused to prevent server-side request forgery.
Sandboxed previews
HTML prototypes produced by the agent are previewed in a sandboxed frame. They can run their own scripts, but they cannot reach your session, cookies or the workbench around them.
Logs without content
Operational logs record request times, error codes and usage counts so we can run the service. They never contain prompts, model responses, deliverable content, API keys or session tokens.
We do not use third-party analytics or advertising trackers on the website or in the workbench.
HTTPS and secure sessions
All traffic is served over HTTPS, and the workbench sends HTTP Strict Transport Security. You stay signed in with a single session cookie that is HttpOnly, SameSite=Lax and Secure.
Passwords are stored only as Argon2id hashes, and sign-in attempts are rate-limited.
People approve the work
Machine checks never mark work as complete. Every deliverable waits for a person to approve it, and improvements the agent proposes to its own rules and skills take effect only after an admin approves them.
Your model providers
When the agent runs, the content it needs is sent to the model provider you configured, using your own API key. That provider processes it under its own terms. We do not use your content to train models.
Limits during Beta
Some account features are not available yet. Please plan for them:
- Password reset is not available. Keep your password somewhere safe, such as a password manager.
- Two-factor authentication and single sign-on are not available yet.
- Account and workspace deletion is not self-service yet.
Give your next project an agent that follows your rules
Sign up with any account name and a password, connect a model provider and submit your first requirement in a few minutes.
Free during Beta · No credit card required