tbh-secure-agents 0.2.2

A secure multi-agent framework by TBH.AI with enhanced security features, guardrails, and protection against AI attacks.

TBH Secure Agents

A secure multi-agent framework by TBH.AI with enhanced security features, guardrails, and protection against AI attacks.

Key Features

• Enhanced Security: Built-in protection against prompt injection, data leakage, and other AI security threats
• Guardrails: Dynamic control of expert behavior through template variables and conditional formatting
• Security Profiles: Predefined security configurations for different use cases
• Multi-Agent Collaboration: Coordinate multiple AI experts to solve complex problems
• Structured Outputs: Ensure consistent and reliable results

Installation

pip install tbh-secure-agents

Documentation

Full documentation, including installation instructions, usage guides, and details on the security features, can be found in the docs/ directory:

• Installation Guide
• Usage Guide
• Security Features
• Guardrails Guide
• Security Profiles
• Version Changes

Quick Start

from tbh_secure_agents import Expert, Operation, Squad
import os

# Set your API key
api_key = os.environ.get('GOOGLE_API_KEY')

# Create experts with security profiles
researcher = Expert(
    specialty="Research Expert specializing in {topic_area}",
    objective="Research and analyze information about {specific_topic}",
    background="You have extensive knowledge in {topic_area} research.",
    security_profile="high_security",
    api_key=api_key
)

writer = Expert(
    specialty="Content Writer",
    objective="Create engaging content based on research findings",
    background="You excel at creating clear, concise content.",
    security_profile="medium_security",
    api_key=api_key
)

# Create operations with template variables
research_operation = Operation(
    instructions="""
    Research the topic of {specific_topic} within the field of {topic_area}.
    Focus on recent developments and key concepts.

    {depth, select,
      basic:Provide a high-level overview suitable for beginners.|
      intermediate:Include more detailed information for those with some knowledge.|
      advanced:Provide in-depth analysis for experts in the field.
    }
    """,
    output_format="A comprehensive research summary with key findings",
    expert=researcher
)

writing_operation = Operation(
    instructions="""
    Based on the research findings, create a {content_type} about {specific_topic}.

    {tone, select,
      formal:Use a professional, academic tone.|
      conversational:Use a friendly, approachable tone.|
      technical:Use precise technical language.
    }

    The content should be suitable for a {audience_level} audience.
    """,
    output_format="A well-structured {content_type} with clear sections",
    expert=writer
)

# Create a squad with the experts and operations
research_squad = Squad(
    experts=[researcher, writer],
    operations=[research_operation, writing_operation],
    process="sequential",
    security_level="high"
)

# Define guardrail inputs
guardrail_inputs = {
    "topic_area": "artificial intelligence",
    "specific_topic": "large language models",
    "depth": "intermediate",
    "content_type": "blog post",
    "tone": "conversational",
    "audience_level": "general"
}

# Deploy the squad with guardrails
result = research_squad.deploy(guardrails=guardrail_inputs)
print(result)

Contributing

Contributions are welcome! Please see the CONTRIBUTING.md file (to be created) and refer to the documentation in the docs/ directory for project structure and goals.

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

Contact

TBH.AI Saish - [e-mail]