# Copyright (c) Microsoft Corporation. # Licensed under the MIT license. """ Core scenario techniques. ``core`` is the home for general-purpose attack techniques usable by any scenario. The ``core`` group tag is injected by ``build_technique_factories`` — factories here carry only their behavioral tags (e.g. ``single_turn``/``multi_turn``/``light``). ``default`` is intentionally not a tag here: what runs by default is scenario-relative and is declared per scenario (see ``AttackTechniqueRegistry.build_technique_class_from_factories``'s ``default_tags`` / ``default_names``), not baked into the shared catalog. """ from pyrit.common.path import ( EXECUTOR_RED_TEAM_PATH, EXECUTOR_SEED_PROMPT_PATH, EXECUTOR_SIMULATED_TARGET_PATH, ) from pyrit.converter import FlipConverter, TaskFramingConverter from pyrit.executor.attack import ( AttackConverterConfig, ManyShotJailbreakAttack, PrependedConversationConfig, PromptSendingAttack, RedTeamingAttack, TreeOfAttacksWithPruningAttack, ) from pyrit.models import AttackTechniqueSeedGroup, SeedPrompt from pyrit.prompt_normalizer import ConverterConfiguration from pyrit.scenario.core.attack_technique_factory import AttackTechniqueFactory def get_technique_factories() -> list[AttackTechniqueFactory]: """ Build the core scenario technique factories. A bare ``PromptSendingAttack`` factory is intentionally omitted: every scenario whose ``BASELINE_ATTACK_POLICY`` is ``BaselineAttackPolicy.Enabled`` already emits an equivalent baseline atomic attack via ``build_baseline_atomic_attack``. Factories that need an adversarial chat target do not bake one in; the default adversarial target is resolved lazily inside ``AttackTechniqueFactory.create`` (for the attack's own ``attack_adversarial_config``) and inside ``AttackTechniqueFactory.resolve_adversarial_chat`` (for the ``AtomicAttack`` that expands a simulated-conversation seed), both via ``get_default_adversarial_target()``. Returns: list[AttackTechniqueFactory]: The core scenario techniques. """ return [ AttackTechniqueFactory.with_simulated_conversation( name="role_play_movie_script", description="Frames the objective as a movie beat by casting the adversarial chat as a screenwriter.", adversarial_chat_system_prompt_path=( EXECUTOR_SEED_PROMPT_PATH / "red_teaming" / "role_play" / "role_play_movie_script.yaml" ), next_message_system_prompt_path=EXECUTOR_SIMULATED_TARGET_PATH / "role_play_next_message.yaml", technique_tags=["single_turn", "light"], num_turns=2, ), AttackTechniqueFactory.with_simulated_conversation( name="role_play_video_game", description="Frames the objective as an in-game mechanic by casting the adversarial chat as a designer.", adversarial_chat_system_prompt_path=( EXECUTOR_SEED_PROMPT_PATH / "red_teaming" / "role_play" / "role_play_video_game.yaml" ), next_message_system_prompt_path=EXECUTOR_SIMULATED_TARGET_PATH / "role_play_next_message.yaml", technique_tags=["single_turn", "light"], num_turns=2, ), AttackTechniqueFactory.with_simulated_conversation( name="role_play_trivia_game", description="Frames the objective as a quiz answer by casting the adversarial chat as a trivia host.", adversarial_chat_system_prompt_path=( EXECUTOR_SEED_PROMPT_PATH / "red_teaming" / "role_play" / "role_play_trivia_game.yaml" ), next_message_system_prompt_path=EXECUTOR_SIMULATED_TARGET_PATH / "role_play_next_message.yaml", technique_tags=["single_turn", "light"], num_turns=2, ), AttackTechniqueFactory.with_simulated_conversation( name="role_play_persuasion", description="Frames a persuasion-resistance test so the target drafts a persuasive spoken script.", adversarial_chat_system_prompt_path=( EXECUTOR_SEED_PROMPT_PATH / "red_teaming" / "role_play" / "role_play_persuasion.yaml" ), next_message_system_prompt_path=EXECUTOR_SIMULATED_TARGET_PATH / "role_play_next_message.yaml", technique_tags=["single_turn", "light"], num_turns=2, ), AttackTechniqueFactory.with_simulated_conversation( name="role_play_persuasion_written", description="Frames a persuasion-resistance test so the target drafts written persuasive material.", adversarial_chat_system_prompt_path=( EXECUTOR_SEED_PROMPT_PATH / "red_teaming" / "role_play" / "role_play_persuasion_written.yaml" ), next_message_system_prompt_path=EXECUTOR_SIMULATED_TARGET_PATH / "role_play_next_message.yaml", technique_tags=["single_turn", "light"], num_turns=2, ), AttackTechniqueFactory( name="many_shot", attack_class=ManyShotJailbreakAttack, description="Primes the target with many fake example exchanges that model compliance before the ask.", technique_tags=["multi_turn", "light"], ), AttackTechniqueFactory( name="tap", attack_class=TreeOfAttacksWithPruningAttack, description="Explores a tree of adversarial prompts, pruning weak branches to refine the attack.", technique_tags=["multi_turn"], ), AttackTechniqueFactory.with_simulated_conversation( name="crescendo_simulated", description="Escalates gradually over a simulated conversation toward the objective.", technique_tags=["single_turn"], ), AttackTechniqueFactory.with_simulated_conversation( name="crescendo_movie_director", description="Escalates gradually while posing as a movie director coaxing the target scene by scene.", technique_tags=["single_turn"], ), AttackTechniqueFactory.with_simulated_conversation( name="crescendo_history_lecture", description="Escalates gradually while framing the ask as an academic history lecture.", technique_tags=["single_turn"], ), AttackTechniqueFactory.with_simulated_conversation( name="crescendo_journalist_interview", description="Escalates gradually while posing as a journalist drawing the target out.", technique_tags=["single_turn"], ), AttackTechniqueFactory( name="red_teaming", attack_class=RedTeamingAttack, description="Uses an adversarial chat model to converse with the target and adapt toward the objective.", technique_tags=["multi_turn", "light"], ), AttackTechniqueFactory.with_simulated_conversation( name="context_compliance", description="Injects a fabricated prior exchange so the target continues as if it already agreed.", adversarial_chat_system_prompt_path=EXECUTOR_RED_TEAM_PATH / "context_compliance" / "context_compliance.yaml", simulated_target_system_prompt_path=EXECUTOR_SIMULATED_TARGET_PATH / "context_compliance_target.yaml", final_user_message="yes.", num_turns=1, technique_tags=["single_turn", "light"], ), AttackTechniqueFactory( name="flip", attack_class=PromptSendingAttack, description="Reverses the objective text so it slips past filters, then asks the target to flip it back.", technique_tags=["single_turn", "light"], attack_kwargs={ "attack_converter_config": AttackConverterConfig( request_converters=ConverterConfiguration.from_converters( converters=[FlipConverter(), TaskFramingConverter(strip_characters="'")] ) ), "prepended_conversation_config": PrependedConversationConfig(apply_converters_to_roles=["user"]), }, seed_technique=AttackTechniqueSeedGroup.from_system_prompt( SeedPrompt.from_yaml_file(EXECUTOR_SEED_PROMPT_PATH / "flip_attack.yaml").value ), ), ]