sherpa_ai.config package

In This Page:

sherpa_ai.config package#

Overview#

The config package provides configuration management tools for Sherpa AI applications. It allows users to define and customize task-specific settings and parameters to control agent behavior.

Key Components

TaskConfig: Configuration class for defining task-specific settings
Configuration Management: Tools for loading and saving configurations
Parameter Validation: Ensures configuration parameters are valid

Example Usage#

from sherpa_ai.config import TaskConfig

# Create a task configuration
config = TaskConfig(
    task_name="Question Answering",
    model_name="gpt-4",
    temperature=0.7,
    max_tokens=1000
)

# Use the configuration with an agent
from sherpa_ai.agents import QAAgent

agent = QAAgent(config=config)
response = agent.get_response("What is artificial intelligence?")

Submodules#

Module	Description
`sherpa_ai.config.task_config`	Provides the TaskConfig class for managing agent and task-specific configurations.

sherpa_ai.config.task_config module#

class sherpa_ai.config.task_config.AgentConfig(**data)[source]#

Bases: BaseModel

Configuration settings for agent behavior and capabilities.

This class defines various configuration options that control how agents operate, including verbosity, search domains, reflection capabilities, and task agent usage.

verbose#

Whether to enable verbose messaging during agent execution. Defaults to True.

Type:: bool

gsite#

List of domains to be used for Google search. Defaults to an empty list.

Type:: list[str]

do_reflect#

Whether to enable the reflection step for each agent. Defaults to False.

Type:: bool

use_task_agent#

Whether to enable use of task agent (obsolete). Defaults to False.

Type:: bool

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig(verbose=True, gsite=["example.com"])
>>> print(config.verbose)
True
>>> print(config.search_domains)
['example.com']

verbose: bool#

gsite: list[str]#

do_reflect: bool#

use_task_agent: bool#

classmethod parse_gsite(value)[source]#

Parse a comma-separated string of URLs into a list.

Parameters:: value (Optional[str]) – A comma-separated string of URLs.
Returns:: A list of stripped URL strings.
Return type:: list[str]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> result = AgentConfig.parse_gsite("example.com, test.com")
>>> print(result)
['example.com', 'test.com']

property search_domains: List[str]#

Get a list of valid search domains.

Returns:: A list of valid URLs from the gsite attribute.
Return type:: List[str]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig(gsite=["https://example.com", "invalid-url"])
>>> print(config.search_domains)
['https://example.com']

property invalid_domains: List[str]#

Get a list of invalid search domains.

Returns:: A list of invalid URLs from the gsite attribute.
Return type:: List[str]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig(gsite=["https://example.com", "invalid-url"])
>>> print(config.invalid_domains)
['invalid-url']

classmethod from_input(input_str)[source]#

Parse input string into AgentConfig.

This method extracts configuration parameters from a string that contains both content and configuration options. Configuration options are expected to be at the end of the string, separated by ‘–‘.

Parameters:

input_str (str) – The input string containing content and configuration.

Returns:

A tuple containing the content part and the: parsed AgentConfig object.

Return type:

Tuple[str, AgentConfig]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> content, config = AgentConfig.from_input("Hello world --gsite example.com")
>>> print(content)
Hello world
>>> print(config.gsite)
['example.com']

classmethod from_config(configs)[source]#

Create an AgentConfig from command-line style arguments.

This method parses a list of command-line style arguments and creates an AgentConfig object with the appropriate settings.

Parameters:: configs (List[str]) – List of command-line style arguments.
Returns:: A new AgentConfig instance with settings from the arguments.
Return type:: AgentConfig
Raises:: ValueError – If invalid configuration options are provided.

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig.from_config(["--concise", "--gsite", "example.com"])
>>> print(config.verbose)
False
>>> print(config.gsite)
['example.com']

model_config: ClassVar[ConfigDict] = {}#: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

sherpa_ai.config.task_config.validate_url(url)[source]#

Check if a string is a valid URL.

This function validates a URL by checking if it has both a scheme and a netloc.

Parameters:: url (str) – The URL string to validate.
Returns:: True if the URL is valid, False otherwise.
Return type:: bool

Example

>>> from sherpa_ai.config.task_config import validate_url
>>> print(validate_url("https://example.com"))
True
>>> print(validate_url("invalid-url"))
False

Module contents#

App configuration settings.

This module provides configuration settings for the Sherpa AI application, loading values from environment variables or a .env file. It includes settings for logging, language models, Slack integration, vector databases, and various API keys.

Usage:

First define variables in runtime environment or in your .env file. See .env-sample file for examples and a useful starting point. Then, in your code, use the values like this:

import Config as cfg secret = cfg.SLACK_SIGNING_SECRET another_variable = cfg.ANOTHER_ENVIRONMENT_VARIABLE

To add, remove, or change variables, … 1. Update this file to create the variables 2. Update env-sample to match 3. Update your own .env file and test the changes 4. Update corresponding secrets in Github and deployment environments

class sherpa_ai.config.AgentConfig(**data)[source]#

Bases: BaseModel

Configuration settings for agent behavior and capabilities.

This class defines various configuration options that control how agents operate, including verbosity, search domains, reflection capabilities, and task agent usage.

verbose#

Whether to enable verbose messaging during agent execution. Defaults to True.

Type:: bool

gsite#

List of domains to be used for Google search. Defaults to an empty list.

Type:: list[str]

do_reflect#

Whether to enable the reflection step for each agent. Defaults to False.

Type:: bool

use_task_agent#

Whether to enable use of task agent (obsolete). Defaults to False.

Type:: bool

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig(verbose=True, gsite=["example.com"])
>>> print(config.verbose)
True
>>> print(config.search_domains)
['example.com']

classmethod from_config(configs)[source]#

Create an AgentConfig from command-line style arguments.

This method parses a list of command-line style arguments and creates an AgentConfig object with the appropriate settings.

Parameters:: configs (List[str]) – List of command-line style arguments.
Returns:: A new AgentConfig instance with settings from the arguments.
Return type:: AgentConfig
Raises:: ValueError – If invalid configuration options are provided.

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig.from_config(["--concise", "--gsite", "example.com"])
>>> print(config.verbose)
False
>>> print(config.gsite)
['example.com']

classmethod from_input(input_str)[source]#

Parse input string into AgentConfig.

This method extracts configuration parameters from a string that contains both content and configuration options. Configuration options are expected to be at the end of the string, separated by ‘–‘.

Parameters:

input_str (str) – The input string containing content and configuration.

Returns:

A tuple containing the content part and the: parsed AgentConfig object.

Return type:

Tuple[str, AgentConfig]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> content, config = AgentConfig.from_input("Hello world --gsite example.com")
>>> print(content)
Hello world
>>> print(config.gsite)
['example.com']

property invalid_domains: List[str]#

Get a list of invalid search domains.

Returns:: A list of invalid URLs from the gsite attribute.
Return type:: List[str]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig(gsite=["https://example.com", "invalid-url"])
>>> print(config.invalid_domains)
['invalid-url']

model_config: ClassVar[ConfigDict] = {}#: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

classmethod parse_gsite(value)[source]#

Parse a comma-separated string of URLs into a list.

Parameters:: value (Optional[str]) – A comma-separated string of URLs.
Returns:: A list of stripped URL strings.
Return type:: list[str]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> result = AgentConfig.parse_gsite("example.com, test.com")
>>> print(result)
['example.com', 'test.com']

property search_domains: List[str]#

Get a list of valid search domains.

Returns:: A list of valid URLs from the gsite attribute.
Return type:: List[str]

Example

>>> from sherpa_ai.config.task_config import AgentConfig
>>> config = AgentConfig(gsite=["https://example.com", "invalid-url"])
>>> print(config.search_domains)
['https://example.com']

verbose: bool#

gsite: list[str]#

do_reflect: bool#

use_task_agent: bool#