A single reference for the rules you’ll encounter across the dashboard.
| Rule |
Value |
| Password |
8–100 characters, at least one letter and one number |
| Email |
One account per email address |
| Session length |
24 hours, then sign in again |
| Group membership |
Each user can be in one group at a time |
| Item |
Rule |
| Group name |
3–50 characters; letters, numbers, underscores, hyphens |
| Collection name |
Letters, numbers, underscores only; must be unique (auto-provisioned collections are named for you) |
| Pipeline name |
1–255 characters |
| Pipeline description |
Up to 1,000 characters |
| Data source: Local System Name |
5–15 characters; letters, numbers, underscores, hyphens |
| SMB port |
1–65535 (default 445) |
| Rule |
Value |
| Maximum file size |
50 MB (larger files skipped) |
| File types |
See supported file types |
| Pipelines per source + collection pair |
1 |
| Concurrent runs per pipeline |
1 (a second start is rejected while one is running) |
| Workspace ingestion queue |
Runs execute one at a time across the workspace |
| Sync model |
Full refresh per run — the target collection is rebuilt from scratch |
| Scheduling |
Manual only; runs are started from the pipeline’s Monitoring tab |
| Rule |
Value |
| Scope |
Questions search the collections your group can access — narrow further per chat with the collection picker |
| Relevance |
Only passages sufficiently similar to your question are used |
| Context budget |
The excerpt context provided to the model is capped, so only the top matches contribute |
| Streaming |
Responses stream in word by word; Stop generation cancels mid-response |
| Citations |
Sources are shown for the session an answer was generated in; they aren’t stored, so a reloaded chat shows the text without its underlines |
