Skip to content

AgentFriendlyOptions

type AgentFriendlyOptions =
    {
        Formats: string list
        MarkdownCopies: bool
        LlmsTxt: string option
        LlmsFullTxt: string option
        AlternateLink: bool
        Summary: string option
        Details: string option
        DescriptionKeys: string list
        Section: Page -> string
        ExcludeCollections: string list
    }
AssemblyPartas.Nacara.Plugins.AgentFriendly

Options of the agent-friendly plugin.

Fields

Formats: string list

Source formats whose pages are published as markdown too, lowercase and dotted.

A page's body is written as it was authored, so only a format that is already markdown reads well that way. Defaults to [ ".md" ].

MarkdownCopies: bool

Write a markdown copy of each page beside its html.

/guide/setup/ is copied to /guide/setup.md, the url an agent guesses. Defaults to true.

LlmsTxt: string option

Where the llms.txt index is written, or None for none.

Defaults to Some "llms.txt".

LlmsFullTxt: string option

Where every markdown page is written as one file, or None for none.

Defaults to Some "llms-full.txt".

AlternateLink: bool

Point each page's head at its markdown copy with <link rel="alternate" type="text/markdown">.

Defaults to true. Needs MarkdownCopies.

Summary: string option

The line quoted under the title of llms.txt.

Defaults to None, which uses the site's description.

Details: string option

Markdown placed after the summary in llms.txt: what an agent should know before it reads any page.

Defaults to None.

DescriptionKeys: string list

The front matter keys a page's description is read from, first match wins.

Defaults to [ "description"; "summary" ].

Section: Page -> string

The heading a page is listed under in llms.txt.

Defaults to the first segment of its route, title-cased, and Pages for a page at the top of the site.

ExcludeCollections: string list

Collections to leave out of everything.

Defaults to [].