LeafMesh — Multi-Agent AI Orchestration Platform

YAML-native multi-agent AI platform with self-healing and evolutionary capabilities

LeafMesh transforms multi-agent AI development through declarative YAML configuration that becomes executable intelligence. Built on the MANAGED_MESH architecture with production-ready coordination and persistence.

Core Features

YAML-Native Intelligence - Zero-code agent creation with AST-parsed configuration
Built-in Coordination - Manager and Summarizer agents provide automatic oversight
MANAGED_MESH Architecture - Direct agent communication with conditional routing
Redis-Powered Persistence - Automatic session management and conversation history
Enterprise Tool Ecosystem - 15+ built-in tools with OpenAI-compatible function calling
Advanced Parallel Processing - Multi-session threading with intelligent coordination

Production Features

Self-Healing Networks - 6 autonomous healing actions with failure detection
Evolutionary Optimization - Genetic algorithms with real fitness testing
Adaptive Model Intelligence - ML-powered model selection with performance prediction

Quick Start

1. Installation

pip install leafmesh

2. Environment Setup

# Required: OpenAI API key
export OPENAI_API_KEY="your-openai-key"

# Optional: Additional providers
export ANTHROPIC_API_KEY="your-anthropic-key"
export GOOGLE_API_KEY="your-google-key"

3. Redis Setup

Local Redis:

# macOS
brew install redis && brew services start redis

# Ubuntu/Debian
sudo apt install redis-server && sudo systemctl start redis

# Docker
docker run -d -p 6379:6379 redis:alpine

4. Basic Usage

from leafmesh import LeafMesh

# Initialize from YAML configuration
sdk = LeafMesh.from_yaml("config.yaml")

# Start the mesh
await sdk.start()

# Process requests
response = await sdk.process_request(
    session_id="user_session",
    input_data={"message": "Hello, how can you help me?"}
)

print(response)

Example YAML Configuration:

name: "my_mesh"
architecture: "managed_mesh"

# Built-in coordination
manager:
  enabled: true
  model: "gpt-4o"

summarizer:
  enabled: true
  model: "gpt-4o-mini"

# User-defined agents
agents:
  conversation_agent:
    name: "conversation_agent"
    model: "gpt-4o-mini"
    prompt: "You are a helpful AI assistant."
    yields:
      response: "string"
      confidence: "number"
    tools: ["calculator", "current_time"]

Architecture Overview

LeafMesh implements a MANAGED_MESH architecture with:

LLM Agents - YAML-defined with optional Python enhancement
Manager Agent - Built-in coordination and rule enforcement
Summarizer Agent - Omnipresent monitoring and analysis
Redis Persistence - Automatic session and conversation storage
Event System - All communication flows through events
Tool System - OpenAI-compatible function calling

For detailed architecture information, see docs/ARCHITECTURE.md

Agent Enhancement

Add Python logic to YAML-defined agents:

@sdk.intelligence("conversation_agent")
async def enhance_conversation(llm_response, input_data, context):
    """Add business logic to agent responses"""

    # Access conversation history
    history = context.get("conversation_history", [])

    # Enhance the LLM response
    enhanced_response = add_context(llm_response, history)

    # Trigger other agents conditionally
    if needs_specialist(enhanced_response):
        await sdk.trigger_agents(data={"analysis": enhanced_response})

    return {
        "response": enhanced_response,
        "confidence": calculate_confidence(enhanced_response)
    }

Revolutionary Features

Self-Healing Networks

# Enable automatic failure recovery
await sdk.enable_self_healing()

# Monitor agent health
health = await sdk.get_agent_health_status()
stats = await sdk.get_healing_statistics()

Evolutionary Optimization

# Optimize mesh configuration automatically
test_scenarios = [
    {"input": "Test case 1", "agents": ["conversation_agent"]},
    {"input": "Test case 2", "agents": ["technical_agent"]}
]

best_genome = await sdk.evolve_swarm_architecture(test_scenarios)
await sdk.apply_evolved_configuration()

Adaptive Model Selection

# Automatic model selection based on request characteristics
response = await sdk.adaptive_execute(
    prompt="Analyze this complex scenario",
    preferred_models=["gpt-4o", "claude-3.5-sonnet"]
)

Documentation

Architecture Guide - Technical implementation details
Debugging Guide - Troubleshooting and monitoring
Getting Started - Run create-leafmesh my-project to scaffold a complete example project

Use Cases

LeafMesh excels at:

Customer Service Systems - Multi-tier workflows with self-healing
Data Analysis Pipelines - Collaborative analytical workflows
Content Creation - Coordinated writing and editing
Decision Support - Complex decision-making with oversight
Workflow Automation - Business process automation

Framework Comparison

Feature	LeafMesh	LangGraph	CrewAI	AutoGen
YAML Configuration	Primary	Code-based	Code-based	Code-based
Built-in Coordination	Manager/Summarizer	Manual	Manual	Manual
Auto-Persistence	Redis	Manual	Manual	Manual
Self-Healing	Production	None	None	None
Evolutionary Optimization	Genetic Algorithm	None	None	None

Licensing

LeafMesh is commercial software owned by LeafCraft.

Evaluation: 30-day free evaluation for research/development
Commercial: Requires valid commercial license for revenue-generating use
Enterprise: Custom enterprise licensing available

Licensing: info@leafcraftstudios.com

Getting Started

Install LeafMesh and set up Redis
Create your first YAML configuration with basic agents
Add Python enhancements for custom logic
Enable revolutionary features for production

LeafMesh: Production-ready multi-agent AI with YAML-driven simplicity

postgres crewai langgraph autogen a2a mcp composio zapier claude-agent openai-agents scheduler slack telegram teams discord kafka sqs mqtt imap listeners email channels knowledge-mongo knowledge-postgres knowledge-pinecone knowledge-docs knowledge connectors dev

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

leafmesh-2.4.103-cp314-none-any.whl (2.9 MB view details)

Uploaded Jul 22, 2026 CPython 3.14

leafmesh-2.4.103-cp313-none-any.whl (2.7 MB view details)

Uploaded Jul 22, 2026 CPython 3.13

leafmesh-2.4.103-cp312-none-any.whl (2.7 MB view details)

Uploaded Jul 22, 2026 CPython 3.12

leafmesh-2.4.103-cp311-none-any.whl (2.8 MB view details)

Uploaded Jul 22, 2026 CPython 3.11

leafmesh-2.4.103-cp310-none-any.whl (1.6 MB view details)

Uploaded Jul 22, 2026 CPython 3.10

File details

Details for the file leafmesh-2.4.103-cp314-none-any.whl.

File metadata

Download URL: leafmesh-2.4.103-cp314-none-any.whl
Upload date: Jul 22, 2026
Size: 2.9 MB
Tags: CPython 3.14
Uploaded using Trusted Publishing? No
Uploaded via: twine/6.2.0 CPython/3.14.2

File hashes

Hashes for leafmesh-2.4.103-cp314-none-any.whl
Algorithm	Hash digest
SHA256	`ccdb3d9a486af0872865c8d9ba3a964e56b5a55dd5ca27781f41741a30caf431`
MD5	`8df1a12d2237eaf702d8f73c0b6bcc46`
BLAKE2b-256	`7ea27234e1154008bfd084b2731f95f24eccad133fd9d22deccce68e2bba85fc`

See more details on using hashes here.

File details

Details for the file leafmesh-2.4.103-cp313-none-any.whl.

File metadata

Download URL: leafmesh-2.4.103-cp313-none-any.whl
Upload date: Jul 22, 2026
Size: 2.7 MB
Tags: CPython 3.13
Uploaded using Trusted Publishing? No
Uploaded via: twine/6.2.0 CPython/3.14.2

File hashes

Hashes for leafmesh-2.4.103-cp313-none-any.whl
Algorithm	Hash digest
SHA256	`34f26f3c0ed73f34d487a93989db1726d9f83f69205871cfb47cbd8118f21f87`
MD5	`4ae90e28b928088b850b37bf38de356b`
BLAKE2b-256	`4c509839c527c250ac8223fbe211ae63a329d0ca32142a8a29a6fc823b5f5109`

See more details on using hashes here.

File details

Details for the file leafmesh-2.4.103-cp312-none-any.whl.

File metadata

Download URL: leafmesh-2.4.103-cp312-none-any.whl
Upload date: Jul 22, 2026
Size: 2.7 MB
Tags: CPython 3.12
Uploaded using Trusted Publishing? No
Uploaded via: twine/6.2.0 CPython/3.14.2

File hashes

Hashes for leafmesh-2.4.103-cp312-none-any.whl
Algorithm	Hash digest
SHA256	`34f9797ad5f11d8e48324833062a0771f46274bdad1473a996e250e642a82e29`
MD5	`1d0b8bbebb5bcd5a93729ad1ab897d5d`
BLAKE2b-256	`2f86bd014cb670cb23d14490637219b0d2d5a681c3915bc7b9e763429e955029`

See more details on using hashes here.

File details

Details for the file leafmesh-2.4.103-cp311-none-any.whl.

File metadata

Download URL: leafmesh-2.4.103-cp311-none-any.whl
Upload date: Jul 22, 2026
Size: 2.8 MB
Tags: CPython 3.11
Uploaded using Trusted Publishing? No
Uploaded via: twine/6.2.0 CPython/3.14.2

File hashes

Hashes for leafmesh-2.4.103-cp311-none-any.whl
Algorithm	Hash digest
SHA256	`bfaa9bf11f14ec30d480913db4932c5ded24790491a3c2fd81c9e9faec2281f4`
MD5	`04d0efce84395ce031afa0a6fe85a731`
BLAKE2b-256	`b517b1fd5a53106291342edce0b44de266ab3ae81ff8f04ec8356c61df7eccd6`

See more details on using hashes here.

File details

Details for the file leafmesh-2.4.103-cp310-none-any.whl.

File metadata

Download URL: leafmesh-2.4.103-cp310-none-any.whl
Upload date: Jul 22, 2026
Size: 1.6 MB
Tags: CPython 3.10
Uploaded using Trusted Publishing? No
Uploaded via: twine/6.2.0 CPython/3.14.2

File hashes

Hashes for leafmesh-2.4.103-cp310-none-any.whl
Algorithm	Hash digest
SHA256	`45da5bd34f68d766cbfaa418de1bfd95304905c1d84b6026834974fdef945394`
MD5	`a6461c80628b26431ec92cb2f742ef4a`
BLAKE2b-256	`c2d1fd137ea2ffab01b0478be5c8bf42296e1f370def8a7f541ec894716d1aac`