Skip to content

FEAT: Support paths in Jailbreak scenario, safe outputs in console printer, file output in MD printer#1635

Open
fdubut wants to merge 3 commits intomicrosoft:mainfrom
fdubut:jailbreak_updates
Open

FEAT: Support paths in Jailbreak scenario, safe outputs in console printer, file output in MD printer#1635
fdubut wants to merge 3 commits intomicrosoft:mainfrom
fdubut:jailbreak_updates

Conversation

@fdubut
Copy link
Copy Markdown
Contributor

@fdubut fdubut commented Apr 21, 2026

Description

Three separate quality of life changes:

  • Add support for arbitrary template paths in the Jailbreak scenario (currently supports only from the template library).
  • Add support for "safe outputs" in the console printer (reduce cognitive load for harmful images).
    • New optional parameter safe_outputs in the constructor, defaults to False (no breaking change).
  • Add support for file output in the Markdown printer.
    • Design choices: outputs to either file or standard output (not both), always append to file (no overwrite).

Tests and Documentation

Added some tests for the Jailbreak scenario and the Markdown printer. No test added to the console printer.

@fdubut fdubut changed the title FEAT: Support paths in Jailbreak scenario, add support for safe outputs in console printer FEAT: Support paths in Jailbreak scenario, safe outputs in console printer, file output in MD printer Apr 21, 2026
num_attempts (Optional[int]): Number of times to try each jailbreak.
jailbreak_names (Optional[List[str]]): List of jailbreak names from the template list under datasets.
to use.
jailbreak_names (Optional[List[str]]): List of jailbreak names from the template list under datasets
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We may want to hold off on the jailbreak changes.

We're grabbing techniques a different way that would be more extensible. But we should use this to do the same thing.

We can do this @fdubut; we can update it here or elsewhere. But we should use #1622

@rlundeen2 rlundeen2 self-assigned this Apr 22, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants