Skip to content

Installation

AgentFlow requires Python 3.11 or later.

Core Package

The core package includes config loading, routing, workflow execution, session management, and the tool registry. It does not include any LLM provider SDKs.

pip install gittielabs-agentflow

Core dependencies are minimal: pydantic, python-frontmatter, pyyaml, and httpx.

Provider Extras

Install the extras for the LLM providers you plan to use:

pip install "gittielabs-agentflow[anthropic]"

Installs anthropic>=0.42.0. Use with AnthropicProvider.

pip install "gittielabs-agentflow[openai]"

Installs openai>=1.0.0. Use with OpenAICompatProvider. Works with any OpenAI-compatible API (OpenAI, Azure OpenAI, Ollama, vLLM, etc.).

pip install "gittielabs-agentflow[google]"

Installs google-genai>=1.0.0. Use with GoogleGenAIProvider.

Optional Extras

Extra Install command What it adds
s3 pip install "gittielabs-agentflow[s3]" S3 storage backend (boto3)
vector pip install "gittielabs-agentflow[vector]" Vector memory with Qdrant (qdrant-client)
orchestration pip install "gittielabs-agentflow[orchestration]" Multi-step orchestration primitives
telemetry pip install "gittielabs-agentflow[telemetry]" Langfuse event handler (langfuse)

Install Everything

pip install "gittielabs-agentflow[all]"

This installs all provider SDKs and optional extras.

Development Setup

Clone the repository and install in editable mode with dev dependencies:

git clone https://github.com/GittieLabs/agentflow.git
cd agentflow
pip install -e ".[dev]"

Dev dependencies include pytest, pytest-asyncio, and ruff for linting.

Run the test suite:

pytest

Run the linter:

ruff check src/ tests/

Environment Variables

Each provider reads its API key from standard environment variables:

Provider Environment Variable
Anthropic ANTHROPIC_API_KEY
OpenAI OPENAI_API_KEY
Google Gemini GOOGLE_API_KEY
Langfuse LANGFUSE_PUBLIC_KEY, LANGFUSE_SECRET_KEY

Set these before running your agents:

export ANTHROPIC_API_KEY="sk-ant-..."

Verifying the Installation

import agentflow
print(agentflow.__version__)
# 0.5.0