SecOps-7/MikroDash v0.8.61

on GitHub

AI Agent answers big questions, and Address Lists handles large blocklists

New

• Token budget in Settings → AI Agent, beside Timeout. How much the model may write
  in one reply, its reasoning included. Default 8,192; raise it if answers come back cut
  off.
• Address Lists opens as one row per list, with its entry, dynamic and disabled
  counts. Click a list to see its entries: up to 500 at a time, with a search that runs
  on the server. A synced blocklist of tens of thousands of entries no longer resends
  every entry to the browser each minute.

Changed

• The Home, Standard and Advanced presets include the new pages, in Settings →
  Visible Pages and in a role's page access. Standard now adds Tools, IP Addresses,
  Address Lists, Interface Lists, IP Pools and DHCP Servers; Advanced includes every page.
• The assistant reads address lists one list at a time: it sees each list and its
  counts, then asks for the list it needs.

Fixed

• The AI Agent no longer gives a blank answer. A reasoning model could spend its
  whole reply budget thinking and write nothing; the budget is larger now, a reply that
  still runs out says so, and one cut short is marked as cut off.
• Unanswered questions are no longer replayed to the model with later ones.
• A role given the Advanced preset can see the new pages, Tools and the AI Agent. It
  gave them no access before.
• Saving, deleting or undoing a row reads only that row from the router, not the
  whole table. Editing one address-list entry among 37,000 took about 12 seconds of
  router reads; it now takes a fraction of a second.

Internal

• A generated table can declare GroupBy; each generated page declares its preset tier.
• The router adapter passes back the id an add reports (Cmd.Ret).