Evalkeeper
    Preparing search index...

    Function promptfooBurpAttacks

    • Reads promptfoo's red team tests written as Burp payloads (promptfoo redteam generate -o tests.burp): one URL-encoded attack per line. They carry no canary, so only your suite's own checks and secrets grade them. Note that promptfoo writes these with a model, on its hosted service unless you set it up otherwise, and sends it your app's description.

      Parameters

      • text: string

        The .burp file's text.

      • source: string = "promptfoo"

        Where it came from, with promptfoo's version.

      • kind: "direct" | "indirect" | "extract" | "hijack" | "lookalike" = "direct"

        What the plugin you generated with tests; default "direct".

      Returns ImportedAttacks