effgen 0.0.1

A comprehensive framework for building agents with Small Language Models

effGen: Enabling Small Language Models as Capable Autonomous Agents

Website | Documentation | Paper | PyPI

---

What is effGen?

effGen is a framework that makes Small Language Models (1B-7B parameters) work as powerful AI agents. While most agent frameworks require massive LLMs, effGen is optimized from the ground up for efficient, smaller models.

from effgen import Agent, load_model

# Create an agent with any model
agent = Agent(
    model=load_model("Qwen/Qwen2.5-3B-Instruct"),
    tools=["calculator", "web_search", "code_executor"]
)

# Run tasks
result = agent.run("Search for the latest AI news and summarize the top 3 stories")
print(result.output)

---

Installation

From PyPI (Recommended)

pip install effgen

With vLLM for faster inference

pip install effgen[vllm]

From Source

git clone https://github.com/ctrl-gaurav/effGen.git
cd effGen

# Option 1: Quick install (recommended)
./install.sh

# Option 2: Quick install for CI (no animations)
./install.sh --quick

# Option 3: Full install (includes vLLM, dev tools)
./install.sh --full

# Option 4: Manual install
pip install -e .

---

Quick Start

CLI Usage

# Run a task
effgen run "What is the capital of France?"

# Interactive chat
effgen chat

# Start API server
effgen serve --port 8000

Python API

from effgen import Agent
from effgen.core.agent import AgentConfig

# Configure your agent
config = AgentConfig(
    name="MyAgent",
    model="Qwen/Qwen2.5-3B-Instruct",
    tools=["calculator", "web_search"],
    enable_memory=True
)

# Create and run
agent = Agent(config)
result = agent.run("Calculate 15% tip on $85.50")

---

Features

Feature	Description
SLM Optimized	Prompt engineering and techniques designed for 1B-7B models
Multi-Model	Supports HuggingFace, OpenAI, Anthropic, Gemini, vLLM
Tool Integration	Built-in tools + MCP, A2A, ACP protocol support
Task Decomposition	Automatic breakdown of complex tasks
Multi-Agent	Coordinate multiple specialized agents
Memory Systems	Short-term and long-term memory
Sandboxed Execution	Safe code execution with Docker

---

Examples

See the examples/ directory:

python examples/basic_agent.py   # Calculator agent with tools
python examples/web_agent.py     # Web search agent

---

Security

effGen provides secure execution options:

• Docker Sandbox: Isolated code execution
• Input Validation: Automatic sanitization
• Rate Limiting: Configurable limits on tool usage

For security policies and vulnerability reporting, see SECURITY.md.

---

Citation

@software{effgen2026,
  title = {effGen: Enabling Small Language Models as Capable Autonomous Agents},
  author = {Srivastava, Gaurav and Hussain, Aafiya and Wang, Chi and Lin, Yingyan and Wang, Xuan},
  year = {2026},
  url = {https://github.com/ctrl-gaurav/effGen},
  version = {0.0.1}
}

---

Links

• Website: effgen.org
• Documentation: effgen.org/docs
• PyPI: pypi.org/project/effgen
• Issues: GitHub Issues

---

License

MIT License - see LICENSE for details.

---

Get Started | Examples | GitHub

Made with care for the AI community