LLM Client¶

The LLM Client module handles communication with language models.

Overview¶

The LLM client provides an abstraction layer for different AI providers, currently supporting OpenAI's GPT models.

Module Structure¶

Located in: alpha_bot/llm/

base.py - Abstract base class
openai_client.py - OpenAI implementation
mock.py - Demo mode simulation

Base Interface¶

from abc import ABC, abstractmethod

class LLMClient(ABC):
    """Abstract interface for LLM providers"""

    @abstractmethod
    def generate_command(self, task: str, context: dict) -> str:
        """Generate shell command from task description"""
        pass

    @abstractmethod
    def analyze_safety(self, command: str) -> dict:
        """Analyze command for safety concerns"""
        pass

    @abstractmethod
    def is_task_complete(self, task: str, results: list) -> bool:
        """Determine if task is complete"""
        pass

OpenAI Client¶

Initialization¶

from alpha_bot.llm import OpenAIClient

client = OpenAIClient(
    api_key="your-api-key",
    model="gpt-4",
    base_url="https://api.openai.com/v1"  # Optional
)

Methods¶

`generate_command(task, context)`¶

Generate a shell command from natural language task.

Parameters: - task (str): Natural language task description - context (dict): Execution context including history

Returns: - str: Shell command to execute

Example:

command = client.generate_command(
    task="list all Python files",
    context={"cwd": "/home/user/project"}
)
# Returns: "find . -name '*.py'"

`analyze_safety(command)`¶

Analyze a command for potential dangers.

Parameters: - command (str): Shell command to analyze

Returns: - dict: Safety analysis with risk level and explanation

Example:

analysis = client.analyze_safety("rm -rf /tmp/*")
# Returns: {
#     "is_dangerous": True,
#     "risk_level": "high",
#     "explanation": "Recursive deletion without confirmation",
#     "alternatives": ["rm -ri /tmp/*", "mv /tmp/* ~/.Trash/"]
# }

Mock Client (Demo Mode)¶

For testing without API calls:

from alpha_bot.llm import MockClient

client = MockClient()
command = client.generate_command("list files", {})
# Returns simulated responses

Configuration¶

Environment Variables¶

OPENAI_API_KEY=your-api-key-here
OPENAI_API_BASE=https://api.openai.com/v1  # Optional
MODEL_NAME=gpt-4  # Optional, defaults to gpt-4

Custom Provider¶

To add a new provider:

from alpha_bot.llm.base import LLMClient

class CustomProvider(LLMClient):
    def __init__(self, api_key: str):
        self.api_key = api_key

    def generate_command(self, task: str, context: dict) -> str:
        # Your implementation
        pass

    def analyze_safety(self, command: str) -> dict:
        # Your implementation
        pass