Skip to main content

Access & Limits

Decide who may use the AI Assistant and cap how much it can spend: answers per user, and tokens for the whole site each month. A debugging view of the prompts sits here too.

Where: Foxiz > AI Assistant > Access & Limits, the last section before Save Changes.

The Access & Limits section: Minimum Role on Author, Requests per User (10 min) on 40, Monthly Token Limit on 0, the note on using fewer tokens, and the Prompt Inspector switch off

Minimum Role​

Users below this role never see the assistant.

  • Choices: Contributor, Author, Editor and Administrator. Default: Author.
  • A custom role counts as the built in role whose permissions it has:
    • Contributor: anyone who can edit posts.
    • Author: anyone who can publish posts.
    • Editor: anyone who can edit other people's posts.
    • Administrator: anyone who can manage the site's options.
  • A user also needs permission to edit the post they are on.

Requests per User (10 min)​

How many AI answers each user can ask for in 10 minutes. Default: 40.

  • From 1 to 1000. An empty field keeps the default.
  • Administrators are not limited.
  • Prompts made for a web AI chat have their own limit, three times higher: 120 by default.
  • A writer who reaches it sees "Request limit reached. Try again in" and the minutes left.

Monthly Token Limit​

Stops AI answers in the editor once the site has used this many tokens this month. Default: 0, which means no limit.

  • Up to 1,000,000,000.
  • It counts the whole site, administrators included, and the reader summaries of Summarize with AI in Connector API Mode.
  • It resets on the 1st of each month. Reset usage data in Connection & usage restarts the count at once.
  • Prompt mode keeps working after the limit, because it uses no tokens from the site.

What writers see as the site gets close:

  • From 80% of the limit, the Foxiz AI box in the editor says how much of this month's limit is used.
  • At the limit, a request answers "This month's AI token limit is used up." The writer can still copy the prompt into a web AI chat.

Tokens this month in Connection & usage shows how much of the limit is used.

Using fewer tokens​

  • Start a new chat for unrelated work. Each request carries a short summary of the chat so far. Click New chat in the assistant window when you move on to something else.
  • Set a spending limit with your AI provider. Monthly Token Limit only counts what this site uses. A limit on your provider's own platform is the safest cap of all.

Prompt Inspector​

For debugging. Adds a View prompt link to each answer in the editor, showing the exact prompt sent and its token counts. Off by default.

  • Only administrators see the link, even with the switch on.
  • Prompts are never logged.

Click Save Changes to keep your changes.