pyrit.models.SeedSimulatedConversation

class SeedSimulatedConversation(*, adversarial_chat_system_prompt_path: str | Path, simulated_target_system_prompt_path: str | Path | None = None, next_message_system_prompt_path: str | Path | None = None, num_turns: int = 3, sequence: int = 0, pyrit_version: str | None = None, **kwargs: Any)

Bases: Seed

Configuration for generating a simulated conversation dynamically.

This class holds the paths and parameters needed to generate prepended conversation content by running an adversarial chat against a simulated (compliant) target.

This is a pure configuration class. The actual generation is performed by generate_simulated_conversation_async in the executor layer, which accepts this config along with runtime dependencies (adversarial_chat target, scorer).

The value property returns a JSON serialization of the config for database storage and deduplication.

num_turns

Number of conversation turns to generate.

adversarial_chat_system_prompt_path

Path to the adversarial chat system prompt YAML.

simulated_target_system_prompt_path

Path to the simulated target system prompt YAML. Defaults to the compliant prompt if not specified.

next_message_system_prompt_path

Optional path to the system prompt for generating an additional user message after the simulated conversation. If provided, a single LLM call generates a final user message that attempts to get the target to fulfill the objective in their next response.

__init__(*, adversarial_chat_system_prompt_path: str | Path, simulated_target_system_prompt_path: str | Path | None = None, next_message_system_prompt_path: str | Path | None = None, num_turns: int = 3, sequence: int = 0, pyrit_version: str | None = None, **kwargs: Any) → None

Initialize a SeedSimulatedConversation.

Parameters:

• adversarial_chat_system_prompt_path – Path to YAML file containing the adversarial chat system prompt.

• simulated_target_system_prompt_path – Optional path to YAML file containing the simulated target system prompt. Defaults to the compliant prompt.

• next_message_system_prompt_path – Optional path to YAML file containing the system prompt for generating a final user message. If provided, after the simulated conversation is generated, a single LLM call generates a user message that attempts to get the target to fulfill the objective. Defaults to None (no next message generation).

• num_turns – Number of conversation turns to generate. Defaults to 3.

• sequence – The starting sequence number for generated turns. When combined with static SeedPrompts, this determines where the simulated turns are inserted. Defaults to 0.

• pyrit_version – PyRIT version for reproducibility tracking. Defaults to current version.

• **kwargs – Additional arguments passed to the Seed base class.

Methods

__init__(*, adversarial_chat_system_prompt_path)	Initialize a SeedSimulatedConversation.
compute_hash()	Compute a deterministic hash of this configuration.
from_dict(data)	Create a SeedSimulatedConversation from a dictionary, typically from YAML.
from_yaml_file(file)	Create a new object from a YAML file.
from_yaml_with_required_parameters(...[, ...])	Load a SeedSimulatedConversation from a YAML file and validate required parameters.
get_identifier()	Get an identifier dict capturing this configuration for comparison/storage.
load_simulated_target_system_prompt(*, ...)	Load and render the simulated target system prompt.
render_template_value(**kwargs)	Renders self.value as a template, applying provided parameters in kwargs.
render_template_value_silent(**kwargs)	Renders self.value as a template, applying provided parameters in kwargs.
set_sha256_value_async()	This method computes the SHA256 hash value asynchronously.

Attributes

added_by
data_type	Return the data type for this seed.
dataset_name
description
name
prompt_group_alias
prompt_group_id
sequence_range	Get the range of sequence numbers this simulated conversation will occupy.
source
value_sha256
value
id
harm_categories
authors
groups
date_added
metadata

authors: Sequence[str] | None

compute_hash() → str

Compute a deterministic hash of this configuration.

Returns:

A SHA256 hash string representing the configuration.

date_added: datetime | None

classmethod from_dict(data: Dict[str, Any]) → SeedSimulatedConversation

Create a SeedSimulatedConversation from a dictionary, typically from YAML.

Expected format:

num_turns: 3 adversarial_chat_system_prompt_path: path/to/adversarial.yaml simulated_target_system_prompt_path: path/to/simulated.yaml # optional

Parameters:

data – Dictionary containing the configuration.

Returns:

A new SeedSimulatedConversation instance.

classmethod from_yaml_with_required_parameters(template_path: str | Path, required_parameters: list[str], error_message: str | None = None) → SeedSimulatedConversation

Load a SeedSimulatedConversation from a YAML file and validate required parameters.

Parameters:

• template_path – Path to the YAML file containing the config.

• required_parameters – List of parameter names that must exist.

• error_message – Custom error message if validation fails.

Returns:

The loaded and validated SeedSimulatedConversation.

Raises:

ValueError – If required parameters are missing.

get_identifier() → Dict[str, Any]

Get an identifier dict capturing this configuration for comparison/storage.

Returns:

Dictionary with configuration details.

groups: Sequence[str] | None

harm_categories: Sequence[str] | None

id: uuid.UUID | None

static load_simulated_target_system_prompt(*, objective: str, num_turns: int, simulated_target_system_prompt_path: str | Path | None = None) → str | None

Load and render the simulated target system prompt.

If no path is provided, returns None (no system prompt). Validates that the template has required objective and num_turns parameters.

Parameters:

• objective – The objective to render into the template.

• num_turns – The number of turns to render into the template.

• simulated_target_system_prompt_path – Optional path to the prompt YAML file. If None, no system prompt is used.

Returns:

The rendered system prompt string, or None if no path is provided.

Raises:

ValueError – If the template doesn’t have required parameters.

metadata: Dict[str, str | int] | None

property sequence_range: range

Get the range of sequence numbers this simulated conversation will occupy.

Each turn generates 2 messages (user + assistant), so num_turns generates num_turns * 2 messages. If next_message_system_prompt_path is set, an additional user message is added at the end.

Returns:

A range object representing the sequence numbers.

value: str