Skip to main content
Eval flags mark replies from the bot that had a problem. Your institution uses two kinds:
  • Platform flags: set up by NeuronCx and raised automatically when a reply is graded.
  • Your own flags: flags you add for problems that matter to your institution, for example a scan booked as a home collection.
This page is for admins. To review the conversations that were flagged, see Review Eval Flags.
Only people from your own institution can change your institution’s flags.

​
Open the Eval flags list

1

Open Setup, Flags

Open Setup and click the Flags tile. Two tabs open: Transaction flags and Eval flags. Click Eval flags.
07 Setup Flags Tile

The Flags tile in Setup

2

Read the list

The list shows every eval flag your institution uses. Platform flags come first, then your own.
  • Flag: the name, with its code in grey underneath.
  • What it means: the description the grader uses to decide when to raise it.
  • Severity: High, Medium or Low, with a coloured dot.
  • Status: On or Off.
Next to + Flag you can see how many of your own flags are on, for example “6 of 15 judge flags on”. Only the table scrolls, so the buttons stay in view.The three platform flags are Answer not backed by content, Off-topic answer and Missing content. You cannot rename them, but you can change their severity and switch them on or off.
08 Setup Eval Flags Tab

The Eval flags tab

​
Add your own flag

1

Click + Flag

The new flag window opens.
2

Name the flag

Type a Name between 3 and 60 characters, for example “Scan booked as home collection”.
3

Describe what should be flagged

Under What should be flagged, describe the problem in a sentence or two, and say when it should not be raised. Use between 10 and 500 characters. The counter under the box shows how many you have used.
4

Choose a severity

Pick High, Medium or Low. It starts on Medium.
5

Choose On or Off

Leave the switch On to start using the flag, or turn it Off to save it for later.
6

Click Create

The new flag appears at the end of the list. NeuronCx makes a short code from the name. The code never changes, even if you rename the flag later.
09 New Flag Dialog

The new flag window

Create stays grey until the name and the description are long enough. A red message under the box tells you how many characters are missing.
Red message under What should be flagged with Create greyed out

The description is too short

Do not use @ or ## in the name or the description. The grader reads this text and does not accept these characters.
A new flag only applies to conversations graded from now on. Conversations graded earlier are not checked again. For the grader to look for your own flags, NeuronCx must switch this on for your institution once.

​
Change a flag

Click the pencil icon on the flag’s row.
  • Your own flags: you can change the Name, the description, the Severity and the switch.
  • Platform flags: the name and description are greyed out, with the note “The name and meaning of this flag are set by the platform. You can change its severity and switch it on or off.” You can still change the severity and the switch.
The small grey line “Code: … (does not change)” shows the flag’s code. Click Save. If you changed nothing, the window just closes. Cancel closes it without saving. Severity changes show in Observe straight away, including on conversations that were already flagged.
10 Edit Platform Flag

Editing a platform flag

​
Switch a flag off

There is no delete. To stop using a flag, open it with the pencil icon, switch it Off, and click Save.
  • The flag stops being raised on new conversations.
  • It no longer shows in the counts, lists and filters in Observe.
  • It stays in your list as Off, so you can switch it on again at any time.
You can have at most 15 of your own flags On at the same time, and 40 in total. The grader gets slower and less accurate with more.

​
Flag stages

Severity is High, Medium or Low. Your institution decides how serious each flag is, and the colours are the same everywhere in the console.

​
If something looks wrong

Review Eval Flags

Find flagged conversations and resolve the flags.

Admin Console

Other platform settings for admins.