Documentation Overview (v0.3 Stable)

Official stable documentation for chatflow-agent v0.3.4: a lightweight, async multi-agent framework with native handoffs for Python.

Stable Release Track (v0.3.4)

You are viewing the official stable documentation for chatflow-agent v0.3.4, recommended for production deployments.

Welcome to the official documentation for chatflow-agent, a lightweight, async-first Python framework engineered for building modular conversational agents and multi-agent swarms with native, type-safe handoffs.

Created and led by Gustavo Baranda alongside an international open-source community, chatflow-agent is designed to eliminate framework bloat and vendor lock-in while providing enterprise-grade concurrency, resilient memory, and multi-channel connectivity.

Architecture Highlights

Zero-SDK Lock-In

Connects to LLMs over pure, asynchronous HTTP REST using httpx. No monolithic vendor SDKs or heavy runtime dependencies required.

Native Multi-Agent Handoffs

Declare collaborative swarms. Agents delegate tasks dynamically using synthesized transfer_to_* tools with automatic cycle prevention.

Transactional Resilient Memory

Every turn takes an atomic snapshot. If an error or timeout occurs, state automatically rolls back. SQLite operations run in dedicated thread pools.

Multi-Channel Readiness

Deploy your agents seamlessly to terminal REPLs, Telegram bots, or WhatsApp Business Cloud API with HMAC SHA-256 signature verification.

Release & Versioning Strategy

chatflow-agent strictly adheres to Semantic Versioning (SemVer). The documentation is version-aware:

Stable (0.3.4) Current Stable Release (Default): Recommended for all production deployments. Published on PyPI on October 23, 2026.
Prerelease (0.4.0rc1) Prerelease Candidate 1: Available on PyPI for testing upcoming architecture enhancements, including REST-native Gemini execution and swarm duplicate validation.

# Install stable release (default)
pip install chatflow-agent

# Install prerelease candidate
pip install "chatflow-agent==0.4.0rc1"

Stable-First Documentation

All code snippets and API signatures in this documentation default to the stable v0.3.4 release. Enhancements or breaking changes introduced in v0.4.0rc1 are explicitly highlighted with the Prerelease (0.4.0rc1) badge.

Getting Started Path

Follow our structured onboarding guide to go from installation to a running production agent in minutes:

Installation: Verify your Python environment (>=3.10), install core packages, and configure optional channel extras.
Quickstart: Build your first agent in under 5 minutes, invoke custom tools with @tool, and explore asynchronous execution.
Core Concepts: Learn the foundational architecture: Agent, Runner, Engines, Tools, Memory, and Handoffs.

Documentation Overview (v0.3 Stable)

Architecture HighlightsCopy Anchor Link