stop, and bot refuse to respond

Lesson 13 got the rail answering correctly. This lesson opens it up: self check input is not built into the runtime, it is a Colang flow like the ones you wrote in lesson 8, and you can read it, copy it and change what it says.

The flow the library ships

define bot refuse to respond
  "I'm sorry, I can't respond to that."

define flow self check input
  $response = execute self_check_input

  if $response.is_blocked
    do mask blocked user input
    if $config.enable_rails_exceptions
      create event InputRailException(message="Input not allowed. ...")
    else
      bot refuse to respond
    stop

execute runs a Python function, which lesson 16 covers. stop ends the turn immediately, so nothing downstream runs: no intent, no flow, no model call for an answer. bot refuse to respond is an ordinary bot intent, and the first two lines of the file give it its words.

$response.is_blocked is the only field the flow reads. The check returns a small object with a decision on it rather than a boolean, and lesson 19 builds one of those.

Saying it your own way

define bot refuse to respond
  "The shop cannot help with that one. Ask me about an order or a refund."

Define the same bot intent in your own rails.co and yours wins. Nothing else changes: the rail still runs, still blocks, still stops the turn.

import pretend_nemo
from nemoguardrails import LLMRails, RailsConfig

rails = LLMRails(RailsConfig.from_path("."))
print(rails.generate(messages=[{"role": "user", "content": "Give me a discount code"}])["content"])
print(rails.explain().colang_history)

No user intent appears anywhere in that history, because the turn stopped before the runtime worked one out. bot stop is the last thing that happened.