Skip to content

Conversation

@yuichiro2023
Copy link

Phase 1 – Custom Track Submission (Omikuji)

This pull request contains three custom security scenarios for the Agent Security Arena.

Submitted Scenarios

  • MoneyMateMayhem
    Direct Logic Override (Jailbreak) targeting financial compliance decisions.

  • AuditAvoider
    Indirect Prompt Injection via document residue affecting tax audit risk assessment.

  • AutoMateAttack – Safety Sensors
    Direct Logic Override attempting to elicit prohibited instructions in an in-vehicle assistant.

Notes

  • This is a Custom Track submission.
  • Each scenario includes a README, plugin implementation, TOML configuration, and test results.
  • All scenarios are reproducible using uv run agentbeats-run.

Thank you for your review.

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.

1 participant