Data Inventory & Classification
Status
Owner: Tech lead · Status: draft · Canonical source: Flezi Galaxy Codebase
Details all categories of data processed by the platform, storage locations, retention periods, and encryption safeguards.
Classification Matrix
| Data Category | Storage Location | Classification | Default Retention | Encryption |
|---|---|---|---|---|
| User Accounts (Email, name, hashed password) | PostgreSQL (users) | Confidential | Account lifetime | bcrypt + at-rest encryption |
| Conversations & Messages | PostgreSQL (ai_messages) | Internal | Configurable (default persistent) | AES-256 at-rest |
| Files & Documents | Azure Blob Storage | Internal / Confidential | Until deleted by user | Azure SSE (AES-256) |
| Vector Embeddings | Azure AI Search | Internal | Synced with document lifecycle | Azure Search at-rest encryption |
| Streaming Events | Redis Stream | Transient | Max 24 hours (TTL expiration) | In-transit TLS + memory TTL |
| Sandbox Artifacts | Sandbox VM Disk → Blob | Transient → Durable | Harvested at run completion | Container deleted after session |
| Audit Logs & Traces | MLflow / PostgreSQL | Internal | 90–365 days | Encrypted at-rest |
Existing Controls
- Least-privilege role scoping on all data operations.
- Automated TTL purge on transient Redis buffers and sandbox containers.
Evidence (Files & Configuration)
flezi.app.api.backend/app/database/models/flezi.app.api.backend/app/features/files/storage.py
Limitations & Roadmap
- Self-service automated retention rules for organizational administrators.