Data Management¶
Data Management exposes Bro's local SQLite environment for general data operations and AI asset governance.
Tabs¶
| Tab | Purpose |
|---|---|
| 📥 Import | Import Excel workbooks and register AI assets. |
| 🗂 Browse | Browse selected SQLite tables. |
| 💉 CRUD | Insert, update, and delete records. |
| 📊 Explore | Inspect/paginate table records. |
| 🔎 Filter | Filter selected table data. |
| 🧮 Aggregate | Compute supported aggregate metrics. |
| 📈 Visualize | Render Plotly visualizations. |
| âš™ Admin | Manage schemas, indexes, tables, columns, profiles, and governed assets. |
| 🧠SQL | Run guarded read-only SQL and export results. |
Core tables¶
| Table | Purpose |
|---|---|
chat_history |
Persistent chat history. |
embeddings |
Semantic chunks, source identity, and vectors. |
Prompts |
Prompt templates and persisted Category metadata. |
documents |
Document governance metadata. |
document_chunks |
Governed chunk metadata/content. |
document_embeddings |
Embedding governance metadata. |
images |
Uploaded-image governance metadata. |
Prompt schema¶
The current Prompts contract is:
Data-management operations should not recreate the removed PromptsId or Version fields.
Guarded SQL¶
The SQL console is intentionally read-only. Mutation statements are rejected so ad hoc analysis does not become an uncontrolled schema/data-write path.
AI asset governance¶
Bro can register active:
- documents;
- document chunks;
- embedding metadata;
- uploaded image metadata.
This metadata is distinct from the model's in-memory context and allows local inspection of managed assets.
Identifier handling¶
Table/column administration validates identifiers before composing schema operations. Arbitrary data values continue to use parameterized SQL where applicable.