Security
Data model
The data OpenStar stores and why.
OpenStar stores data needed to personalize repository discovery and Watch.
Data follows the product loop
Stored data should support discovery, Chat, Watch, Memory, security, and product quality. Data that does not serve those paths should be removed or avoided.
Account
User identity, sign-in links, and connected accounts.
Repository corpus
Public repo metadata, sources, and refresh state.
Memory
Taste graph, user repo edges, and preference signals.
Chat and Watch
Sessions, message parts, briefs, tool calls, and runs.
Data groups
| Group | Why it exists |
|---|---|
| User account and identities | Sign in, ownership, and data controls. |
| GitHub connection and public signals | Personalize from public stars and owned repos. |
| Repository metadata and sources | Build Explore, Chat evidence, and Watch briefs. |
| Memory and user repo edges | Explain and tune recommendations. |
| Chat sessions and message parts | Preserve conversation context and tool evidence. |
| Watch items, updates, runs, and tool calls | Generate and debug briefs. |
| AI provider settings and encrypted key state | Enable BYOK without exposing secrets. |
| Analytics events | Measure activation and product quality. |
The MVP focuses on public repository workflows.