pyrit.executor.attack.TAPAttackResult

class TAPAttackResult(conversation_id: str, objective: str, attack_identifier: dict[str, str], last_response: ~pyrit.models.message_piece.MessagePiece | None = None, last_score: ~pyrit.models.score.Score | None = None, executed_turns: int = 0, execution_time_ms: int = 0, outcome: ~pyrit.models.attack_result.AttackOutcome = AttackOutcome.UNDETERMINED, outcome_reason: str | None = None, related_conversations: set[~pyrit.models.conversation_reference.ConversationReference] = <factory>, metadata: ~typing.Dict[str, ~typing.Any] = <factory>)

Bases: AttackResult

Result of the Tree of Attacks with Pruning (TAP) attack strategy execution.

This result includes the standard attack result information with attack-specific data stored in the metadata dictionary.

__init__(conversation_id: str, objective: str, attack_identifier: dict[str, str], last_response: ~pyrit.models.message_piece.MessagePiece | None = None, last_score: ~pyrit.models.score.Score | None = None, executed_turns: int = 0, execution_time_ms: int = 0, outcome: ~pyrit.models.attack_result.AttackOutcome = AttackOutcome.UNDETERMINED, outcome_reason: str | None = None, related_conversations: set[~pyrit.models.conversation_reference.ConversationReference] = <factory>, metadata: ~typing.Dict[str, ~typing.Any] = <factory>) → None

Methods

__init__(conversation_id, objective, ...[, ...])
duplicate()	Create a deep copy of the result.
get_conversations_by_type(conversation_type)	Return all related conversations of the requested type.

Attributes

auxiliary_scores_summary	Get a summary of auxiliary scores from the best node.
best_adversarial_conversation_id	Get the adversarial conversation ID for the best-scoring branch.
executed_turns
execution_time_ms
last_response
last_score
max_depth_reached	Get the maximum depth reached in the attack tree.
nodes_explored	Get the total number of nodes explored during the attack.
nodes_pruned	Get the number of nodes pruned during the attack.
outcome
outcome_reason
tree_visualization	Get the tree visualization from metadata.
conversation_id
objective
attack_identifier
related_conversations
metadata

attack_identifier: dict[str, str]

property auxiliary_scores_summary: Dict[str, float]

Get a summary of auxiliary scores from the best node.

property best_adversarial_conversation_id: str | None

Get the adversarial conversation ID for the best-scoring branch.

conversation_id: str

property max_depth_reached: int

Get the maximum depth reached in the attack tree.

metadata: Dict[str, Any]

property nodes_explored: int

Get the total number of nodes explored during the attack.

property nodes_pruned: int

Get the number of nodes pruned during the attack.

objective: str

related_conversations: set[ConversationReference]

property tree_visualization: Tree | None

Get the tree visualization from metadata.