AgentKit
DocsKitsCLI ReferenceDesktop App
BetaYou are reading docs for the beta channel (3.0.0-beta.15). Features may change before the next stable release.Switch to stable →

eval

ak eval ui

Report opt-in shadow UI regression signals

Use ak eval ui for the operation described below. The command help records its arguments, effects, and options.

Shadow mode calls the external provider only when you have enabled semantic_decision at user scope, granted this consumer explicitly with semantic_decision.consumers.ui-regression set to true, and made the selected provider's credential available (see providers and credentials). The grant is required even when you have no consumers map. See ak eval decision for the opt-in keys.

--output creates the receipt with mode 0600 and fails if the file already exists.

Command contract

Report opt-in shadow UI regression signals

Synopsis

Reads a reviewed compact DOM/a11y/computed/journey capture. Default off makes no provider call. Shadow requires user provider consent, explicit ui-regression consumer grant, and key. Browser failures remain failures; unknown provider results retain the current workflow.

Effects: Optional paid text-only inference; no browser or vision calls. Creates a receipt only with --output. Never uploads image/video or changes test results.

Exit status: 0 report available; inspect deterministic failures and unknown signals 1 invalid input, configuration or output failure

ak eval ui <state.json|-> [flags]

Examples

  ak eval ui capture.json --json
  ak eval ui capture.json --mode shadow --output signals.json --json

Options

  -h, --help             help for ui
      --json             Emit machine-readable JSON (implies --no-interactive)
      --mode string      off or shadow; never changes browser authority (default "off")
      --no-interactive   Disable interactive prompts (CI-safe)
      --output string    Create a new receipt without overwriting
  -q, --quiet            Suppress non-error output on stderr
  -V, --verbose          Extra diagnostic output on stderr (loses to --quiet)
  -y, --yes              Assume yes for all prompts

SEE ALSO

  • ak eval - Run controlled skill and agent evaluations