versionhq 1.1.10.9

LLM orchestration frameworks for model-agnostic AI agents that handle complex outbound workflows

Overview

LLM orchestration frameworks to deploy multi-agent systems with task-based formation.

Visit:

• PyPI
• Github (LLM orchestration framework)
• Use case / Quick demo
• Documentation *Some components are under review.

---

Table of Content

• Key Features
  • Agent formation
• Quick Start
  • Case 1. Solo Agent:
    • Return a structured output with a summary in string.
  • Case 2. Supervising:
• Technologies Used
• Project Structure
• Setup
• Contributing
  • Customizing AI Agents
  • Modifying RAG Functionality
  • Package Management with uv
  • Pre-Commit Hooks
• Trouble Shooting
• Frequently Asked Questions (FAQ)

---

Key Features

Generate mulit-agent systems depending on the complexity of the task, and execute the task with agents of choice.

Model-agnostic agents can handle RAG tools, tools, callbacks, and knowledge sharing among other agents.

Agent formation

Depending on the task complexity, agents can make a different formation.

You can specify which formation you want them to generate, or let the agent decide if you don’t have a clear plan.

	Solo Agent	Supervising	Network	Random
Formation
Usage	A single agent with tools, knowledge, and memory. When self-learning mode is on - it will turn into Random formation.	Leader agent gives directions, while sharing its knowledge and memory. Subordinates can be solo agents or networks.	Share tasks, knowledge, and memory among network members.	A single agent handles tasks, asking help from other agents without sharing its memory or knowledge.
Use case	An email agent drafts promo message for the given audience.	The leader agent strategizes an outbound campaign plan and assigns components such as media mix or message creation to subordinate agents.	An email agent and social media agent share the product knowledge and deploy multi-channel outbound campaign.	1. An email agent drafts promo message for the given audience, asking insights on tones from other email agents which oversee other clusters. 2. An agent calls the external agent to deploy the campaign.

---

Quick Start

Install versionhq package:

pip install versionhq

(Python >= 3.12)

Case 1. Solo Agent:

Return a structured output with a summary in string.

from pydantic import BaseModel
from versionhq.agent.model import Agent
from versionhq.task.model import Task

class CustomOutput(BaseModel):
   test1: str
   test2: list[str]

def dummy_func(message: str, test1: str, test2: list[str]) -> str:
   return f"{message}: {test1}, {", ".join(test2)}"


agent = Agent(role="demo", goal="amazing project goal")

task = Task(
   description="Amazing task",
   pydantic_custom_output=CustomOutput,
   callback=dummy_func,
   callback_kwargs=dict(message="Hi! Here is the result: ")
)

res = task.execute_sync(agent=agent, context="amazing context to consider.")
print(res)

This will return TaskOutput that stores a response in string, JSON dict, and Pydantic model: CustomOutput formats with a callback result.

res == TaskOutput(
   raw="{\\"test1\\": \\"random str\\", \\"test2\\": [\\"item1\\", \\"item2\\"]}",
   json_dict={"test1": "random str", "test2": ["item1", "item2"]},
   pydantic=CustomOutput(test1="random str", test2=["item 1", "item 2"]),
   callback_output="Hi! Here is the result: random str, item 1, item 2",
)

Case 2. Supervising:

from versionhq.agent.model import Agent
from versionhq.task.model import Task, ResponseField
from versionhq.team.model import Team, TeamMember

agent_a = Agent(role="agent a", goal="My amazing goals", llm="llm-of-your-choice")
agent_b = Agent(role="agent b", goal="My amazing goals", llm="llm-of-your-choice")

task_1 = Task(
   description="Analyze the client's business model.",
   response_fields=[ResponseField(title="test1", data_type=str, required=True),],
   allow_delegation=True
)

 task_2 = Task(
   description="Define the cohort.",
   response_fields=[ResponseField(title="test1", data_type=int, required=True),],
   allow_delegation=False
)

team = Team(
   members=[
      TeamMember(agent=agent_a, is_manager=False, task=task_1),
      TeamMember(agent=agent_b, is_manager=True, task=task_2),
   ],
)
res = team.kickoff()

This will return a list with dictionaries with keys defined in the ResponseField of each task.

Tasks can be delegated to a team manager, peers in the team, or completely new agent.

---

Technologies Used

Schema, Data Validation

• Pydantic: Data validation and serialization library for Python.
• Pydantic_core: Core func packages for Pydantic.
• Upstage: Document processer for ML tasks. (Use Document Parser API to extract data from documents)
• Docling: Document parsing

Storage

• mem0ai: Agents' memory storage and management.
• Chroma DB: Vector database for storing and querying usage data.
• SQLite: C-language library to implements a small SQL database engine.

LLM-curation

• LiteLLM: Curation platform to access LLMs

Tools

• Composio: Conect RAG agents with external tools, Apps, and APIs to perform actions and receive triggers. We use tools and RAG tools from Composio toolset.

Deployment

• Python: Primary programming language. v3.13 is recommended.
• uv: Python package installer and resolver
• pre-commit: Manage and maintain pre-commit hooks
• setuptools: Build python modules

---

Project Structure

.
.github
└── workflows/                # Github actions
│
src/
└── versionhq/                # Orchestration frameworks
│     ├── agent/              # Components
│     └── llm/
│     └── task/
│     └── team/
│     └── tool/
│     └── cli/
│     └── ...
│     │
│     ├── db/                 # Storage
│     ├── chroma.sqlite3
│     └── ...
│
└──tests/                     # Pytest
│     └── agent/
│     └── llm/
│     └── ...
│
└── uploads/                  # Local repo to store the uploaded files

---

Setup

1. Install the uv package manager:

   brew install uv
   

2. Install dependencies:

   uv venv
   source .venv/bin/activate
   uv pip sync
   

• In case of AssertionError/module mismatch, run Python version control using .pyenv

  pyenv install 3.12.8
  pyenv global 3.12.8  (optional: `pyenv global system` to get back to the system default ver.)
  uv python pin 3.12.8
  

3. Set up environment variables: Create a .env file in the project root and add the following:

   OPENAI_API_KEY=your-openai-api-key
   LITELLM_API_KEY=your-litellm-api-key
   UPSTAGE_API_KEY=your-upstage-api-key
   COMPOSIO_API_KEY=your-composio-api-key
   COMPOSIO_CLI_KEY=your-composio-cli-key
   

---

Contributing

1. Create your feature branch (git checkout -b feature/your-amazing-feature)

2. Create amazing features

3. Test the features using the tests directory.

   • Add a test function to respective components in the tests directory.
   • Add your LITELLM_API_KEY, OPENAI_API_KEY, COMPOSIO_API_KEY, DEFAULT_USER_ID to the Github repository secrets located at settings > secrets & variables > Actions.
   • Run a test.

     uv run pytest tests -vv --cache-clear
     

   pytest

   • When adding a new file to tests, name the file ended with _test.py.
   • When adding a new feature to the file, name the feature started with test_.

4. Pull the latest version of source code from the main branch (git pull origin main) *Address conflicts if any.

5. Commit your changes (git add . / git commit -m 'Add your-amazing-feature')

6. Push to the branch (git push origin feature/your-amazing-feature)

7. Open a pull request

Optional

• Flag with #! REFINEME for any improvements needed and #! FIXME for any errors.

• Run a React demo app: React demo app to check it on the client endpoint.

  npm i
  npm start
  

  The frontend will be available at http://localhost:3000.

• production use case is available at https://versi0n.io. Currently, we are running alpha test.

Customizing AI Agents

To add an agent, use sample directory to add new project. You can define an agent with a specific role, goal, and set of tools.

Your new agent needs to follow the Agent model defined in the verionhq.agent.model.py.

You can also add any fields and functions to the Agent model universally by modifying verionhq.agent.model.py.

Modifying RAG Functionality

The RAG system uses Chroma DB to store and query past campaign dataset. To update the knowledge base:

1. Add new files to the uploads/ directory. (This will not be pushed to Github.)
2. Modify the tools.py file to update the ingestion process if necessary.
3. Run the ingestion process to update the Chroma DB.

Package Management with uv

• Add a package: uv add <package>
• Remove a package: uv remove <package>
• Run a command in the virtual environment: uv run <command>

• After updating dependencies, update requirements.txt accordingly or run uv pip freeze > requirements.txt

Pre-Commit Hooks

1. Install pre-commit hooks:

   uv run pre-commit install
   

2. Run pre-commit checks manually:

   uv run pre-commit run --all-files
   

Pre-commit hooks help maintain code quality by running checks for formatting, linting, and other issues before each commit.

• To skip pre-commit hooks (NOT RECOMMENDED)

  git commit --no-verify -m "your-commit-message"
  

---

Trouble Shooting

Common issues and solutions:

• API key errors: Ensure all API keys in the .env file are correct and up to date. Make sure to add load_dotenv() on the top of the python file to apply the latest environment values.
• Database connection issues: Check if the Chroma DB is properly initialized and accessible.
• Memory errors: If processing large contracts, you may need to increase the available memory for the Python process.
• Issues related to dependencies: rm -rf uv.lock, uv cache clean, uv venv, and run uv pip install -r requirements.txt -v.
• Issues related to the AI agents or RAG system: Check the output.log file for detailed error messages and stack traces.
• Issues related to Python quit unexpectedly: Check this stackoverflow article.
• reportMissingImports error from pyright after installing the package: This might occur when installing new libraries while VSCode is running. Open the command pallete (ctrl + shift + p) and run the Python: Restart language server task.

---

Frequently Asked Questions (FAQ)

Q. Where can I see if the agent is working?

A. You can find a frontend app here with real-world outbound use cases. You can also test features here using React app.

Q. How do you analyze the customer?

A. We employ soft clustering for each customer.

Q. When should I use a team vs an agent?

A. In essence, use a team for intricate, evolving projects, and agents for quick, straightforward tasks.

Use a team when:

Complex tasks: You need to complete multiple, interconnected tasks that require sequential or hierarchical processing.

Iterative refinement: You want to iteratively improve upon the output through multiple rounds of feedback and revision.

Use an agent when:

Simple tasks: You have a straightforward, one-off task that doesn't require significant complexity or iteration.

Human input: You need to provide initial input or guidance to the agent, or you expect to review and refine the output.