Portal Polish: Tool Details, Safer Token Limits, and Resilient Health
July 15, 2026 · InfraScout Team
No new capability this time — just a round of portal polish aimed at the people who administer InfraScout. Admins can now see exactly which tools a group contains, AI provider connections save cleanly at higher token limits, and the API status banner stops crying wolf after a brief network blip.
See every tool inside a tool group
The admin Tool Groups view told you how many tools a group held, but not which ones. To find out, you had to open a group for editing — and only custom groups could be edited at all, so built-in and external groups stayed opaque.
Each group card is now clickable and opens a read-only tool-details view. It lists every tool in the group with its name, a short description of what it does, and whether the tool is read-only or can make changes. That classification lets you tell at a glance which groups are safe for read-only assessments and which include tools that write. The view works for all three kinds of groups — built-in, custom, and external — so you get the same clear inventory everywhere, without touching a group's configuration.
Safer token limits on AI provider connections
Saving an AI provider connection with a large maximum-tokens value used to fail with a confusing "connection test failed" message that pointed you at your API key and model name — even when both were correct. The real cause was the connection check itself, which no longer trips over a high token limit. Connections that were rejected before now save as expected.
The maximum-tokens field is also stricter about what you type into it. A mistyped or locale-formatted value such as 32.768 — meant as 32768 — used to slip through and get sent to the provider, which would then reject the request. The field now requires a whole number and flags a bad value right in the form, so you catch the mistake before you save rather than after.
API status that recovers the moment you reconnect
When your network dropped and came back, the portal was slow to notice. The "Backend API unavailable" banner and the disconnected indicator in the header could stay red for several minutes after connectivity had actually returned, because the status only refreshed on a slow, fixed schedule. That was especially confusing next to the live connection dot, which turned green again within a second — leaving you looking at a green dot and a red "disconnected" label at the same time.
The status now recovers as soon as connectivity returns. When your device comes back online or you switch back to the InfraScout tab, the portal rechecks immediately if it was showing an error, so the banner and header catch up right away instead of leaving you wondering whether something is broken.
Questions or feedback? Reach us at info@infrascout.cloud.