Security at stella

stella handles privileged legal data. Security is a core design constraint across the product. The controls below are linked to executable evidence in the public repository, so they change with the implementation rather than relying on an unchecked checklist.

Contact security

Our hosted cloud is currently in beta and developing quickly.

Self-hosted deployments run on your own infrastructure under the open-source license. See our Terms.

Controls you can inspect

Each statement maps to a policy control whose file paths and expected implementation markers are checked in CI.

Authorization boundaries

Workspace access is validated before scoped handlers run, and validated workspace identifiers use a distinct safe type inside the API.

View policy and evidence

Private file storage

Stored objects are private, keys are scoped by organization and workspace, and file access uses short-lived signed URLs.

View policy and evidence

Explicit deletion paths

Document deletion removes database records and stored objects; workspace deletion follows an explicit deletion state rather than silently hiding data.

View policy and evidence

Configurable AI providers

AI provider selection is kept behind a provider abstraction instead of being embedded in product business logic.

View policy and evidence

Dependency safeguards

New dependency releases are quarantined before installation, while automated dependency review blocks known high-severity risk.

View policy and evidence

Continuous static analysis

CodeQL and repository security guards run as committed checks, so security regressions surface in review rather than remaining informal process.

View policy and evidence

Report a vulnerability

Please send security reports privately. Our disclosure policy documents what to include and how we handle a report.