Security News
Research
Data Theft Repackaged: A Case Study in Malicious Wrapper Packages on npm
The Socket Research Team breaks down a malicious wrapper package that uses obfuscation to harvest credentials and exfiltrate sensitive data.
A hackable and extendable framework for developing LLM-based multi-agentic systems.
Please note: diskurs is currently under development (meaning early alpha state) and not yet ready for production use.
Diskurs is a hackable and extendable Python framework for developing LLM-based multi-agentic systems to tackle complex workflow automation tasks. It allows developers to set up agent interactions using customizable configurations.
Diskurs requires Python 3.12 or higher.
This project uses Poetry for dependency management. To install Diskurs and its dependencies:
Clone the repository:
git clone https://github.com/agentic-diskurs/diskurs.git
cd diskurs
Install Poetry (if not already installed):
curl -sSL https://install.python-poetry.org | python3 -
For more details, refer to the official Poetry installation guide.
Install dependencies:
poetry install
Build the documentation:
poetry run sphinx-build docs/source docs/build
Initialize the hook folder:
To setup the hook folder to the custom hook folder run setup_hooks.sh
:
./setup_hooks.sh
Below is an example of how to launch Diskurs in your project:
import logging
from pathlib import Path
from dotenv import load_dotenv
from diskurs import create_forum_from_config, DiskursInput
logger = logging.getLogger(__name__)
logger.setLevel(level=logging.INFO)
ticket_content = """
Hello team, I cannot reach http://www.test.com:8080/ from my machine.
I tried to access it around 04.09.2024 at 1 AM. Please can you check what is wrong?
"""
load_dotenv()
diskurs_input = DiskursInput(
user_query=ticket_content,
metadata={"company_id": "3929", "proxy_instance": "academy-prx002-ch-zur-1"},
)
def main(config: Path):
forum = create_forum_from_config(config_path=config, base_path=Path(__file__).parent)
res = forum.ama(diskurs_input)
print(res)
if __name__ == "__main__":
main(Path(__file__).parent / "config.yaml")
Set Up Configuration: Create a config.yaml
file in your project directory. This file contains the necessary configurations for your agents, tools, and dependencies.
Load Environment Variables: Create a .env
file or set environment variables in your system. This includes API keys and other sensitive information required by Diskurs.
Example .env
file:
AZURE_OPENAI_API_KEY=your_azure_openai_api_key
Run the Script: Execute your Python script to start the forum interaction.
poetry run python your_script.py
The config.yaml
file is used to customize your agents, tools, and dependencies. Below is a condensed example configuration to help you set up your own:
Define the language models your agents will use:
llms:
- name: "gpt-4-base"
type: "azure"
modelName: "gpt-4-0613"
endpoint: "https://your-azure-endpoint.openai.azure.com"
apiVersion: "2023-03-15-preview"
apiKey: "${AZURE_OPENAI_API_KEY}"
Define your agents, specifying their names, types, language models, prompts, tools they use, and the agents they interact with:
agents:
- name: "Conductor_Agent"
type: "conductor"
llm: "gpt-4-base"
prompt:
type: "conductor_prompt"
location: "agents/Conductor_Agent"
userPromptArgumentClass: "ConductorUserPromptArgument"
systemPromptArgumentClass: "ConductorSystemPromptArgument"
longtermMemoryClass: "ConductorLongtermMemory"
canFinalizeName: "can_finalize"
topics:
- "Agent_A"
- "Agent_B"
- name: "Agent_A"
type: "multistep"
llm: "gpt-4-base"
prompt:
type: "multistep_prompt"
location: "agents/Agent_A"
systemPromptArgumentClass: "AgentASystemPrompt"
userPromptArgumentClass: "AgentAUserPrompt"
isValidName: "is_valid"
isFinalName: "is_final"
tools:
- "tool_x"
topics:
- "Conductor_Agent"
List the tools your agents will use and implement the tool functions in the specified modulePath
:
tools:
- name: "tool_x"
functionName: "function_x"
modulePath: "tools/custom_tools.py"
configs:
param1: "value1"
param2: "value2"
Specify external dependencies required by the tools:
toolDependencies:
- type: "external_service"
name: "service_x"
url: "http://service-x-url"
port: 8080
Define Language Models: In the llms
section, configure the language models your agents will use. Replace the endpoint
and apiKey
with your own Azure OpenAI details.
Create Agents: In the agents
section, define your agents. Specify their names, types, language models, prompts, tools they use, and the agents they interact with.
Implement Prompts: For each agent, create the prompt templates and argument classes as specified in the prompt
section. These should be located in the paths you provide under location
.
Add Tools: In the tools
section, list the tools your agents will use. Implement the tool functions in the specified modulePath
.
Configure Dependencies: If your tools rely on external services or databases, specify them in the toolDependencies
section.
Environment Variables: Use environment variables for sensitive information like API keys. Reference them in your config.yaml
using ${VARIABLE_NAME}
.
pyproject.toml
):
diskurs
python-dotenv
Contributions are welcome! Please follow these steps:
Fork the repository on GitHub.
git clone https://github.com/agentic-diskurs/diskurs.git
Create a new branch for your feature or bug fix.
git checkout -b feature/your-feature-name
Commit your changes with clear messages.
git commit -m "Add new feature: description"
Push your branch to your forked repository.
git push origin feature/your-feature-name
Open a pull request detailing your changes.
This project is licensed under the MIT License. See the LICENSE file for details.
For questions, issues, or suggestions, please open an issue on the GitHub repository.
Happy coding with Diskurs!
FAQs
A hackable and extendable framework for developing LLM-based multi-agentic systems.
We found that diskurs demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Research
The Socket Research Team breaks down a malicious wrapper package that uses obfuscation to harvest credentials and exfiltrate sensitive data.
Research
Security News
Attackers used a malicious npm package typosquatting a popular ESLint plugin to steal sensitive data, execute commands, and exploit developer systems.
Security News
The Ultralytics' PyPI Package was compromised four times in one weekend through GitHub Actions cache poisoning and failure to rotate previously compromised API tokens.