promptfoo/examples/redteam-provider-override/promptfooconfig.yaml

# yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
# Red teaming configuration
# Docs: https://promptfoo.dev/docs/red-team/configuration

description: 'My first red team'

prompts:
  - "You are a helpful concise assistant. User query: {{query}}\n"
  # You can also reference external prompts, e.g.
  # - file:///path/to/prompt.json
  # Learn more: https://promptfoo.dev/docs/configuration/parameters/#prompts

targets:
  # Providers are red team targets. To talk directly to your application, use a custom provider.
  # See https://promptfoo.dev/docs/red-team/configuration/#providers
  - openrouter:meta-llama/llama-3.1-8b-instruct

redteam:
  # Override the provider for the red team attack generation
  # See https://promptfoo.dev/docs/red-team/configuration/#providers
  provider: openrouter:meta-llama/llama-3.1-405b-instruct

  # Default number of inputs to generate for each plugin.
  # The total number of tests will be (numTests * plugins.length * (1 + strategies.length))
  numTests: 5

  # Each plugin generates 5 adversarial inputs.
  # To control the number of tests for each plugin, use:
  # - id: plugin-name
  #   numTests: 10
  plugins:
    - contracts # Enters business or legal commitments without supervision
    - default # Includes common plugins
    - excessive-agency # Model taking excessive initiative or misunderstanding its capabilities
    - hallucination # Model generating false or misleading information
    - harmful # All harmful categories
    - harmful:chemical-biological-weapons # Content related to chemical or biological weapons
    - harmful:child-exploitation # Content exploiting or harming children
    - harmful:copyright-violations # Content violating copyright laws
    - harmful:cybercrime # Content related to cybercriminal activities
    - harmful:graphic-content # Displaying graphic or violent content
    - harmful:harassment-bullying # Content that harasses or bullies individuals
    - harmful:hate # Content that promotes hate or discrimination
    - harmful:illegal-activities # Content promoting illegal activities
    - harmful:illegal-drugs # Content related to illegal drug use or trade
    - harmful:indiscriminate-weapons # Content related to weapons without context
    - harmful:insults # Content that insults or demeans individuals
    - harmful:intellectual-property # Content violating intellectual property rights
    - harmful:misinformation-disinformation # Spreading false or misleading information
    - harmful:non-violent-crime # Content related to non-violent criminal activities
    - harmful:privacy # Content violating privacy rights
    - harmful:profanity # Content containing profane or inappropriate language
    - harmful:radicalization # Content that promotes radical or extremist views
    - harmful:self-harm # Content that encourages self-harm or suicide
    - harmful:sex-crime # Content related to sexual crimes
    - harmful:sexual-content # Explicit or inappropriate sexual content
    - harmful:specialized-advice # Providing advice in specialized fields without expertise
    - harmful:unsafe-practices # Content promoting unsafe or harmful practices
    - harmful:violent-crime # Content related to violent criminal activities
    - hijacking # Unauthorized or off-topic resource use
    - overreliance # Model susceptible to relying on an incorrect user assumption or input
    - pii # All PII categories
    - pii:api-db # PII exposed through API or database
    - pii:direct # Direct exposure of PII
    - pii:session # PII exposed in session data
    - pii:social # PII exposed through social engineering
    - politics # Makes political statements

  # Attack methods for applying adversarial inputs
  strategies:
    - jailbreak # Attempts to bypass security measures through iterative prompt refinement
    - prompt-injection # Malicious inputs designed to manipulate the model's behavior