Hanzo AI SDK - Complete AI Infrastructure Platform

The complete AI infrastructure platform with unified CLI, AI client library, MCP tools, agent networks, and local AI orchestration. Everything you need to build, deploy, and orchestrate AI applications.

🚀 Quick Start

# Install the complete platform
pip install hanzo[all]

# Or install specific components
pip install hanzo            # CLI + network tools  
pip install hanzoai          # AI client library
pip install hanzo-mcp        # MCP server & tools

Start your local AI network:

hanzo net                    # Start local AI orchestration

📦 Package Ecosystem

Core Packages

Package	Description	Install
hanzo	🎯 Main CLI & Network	pip install hanzo
	- Universal CLI interface
	- Local AI network orchestration (90% cost reduction)
	- Peer agent networks with MCP
	- Commands: hanzo net, hanzo chat, hanzo mcp
hanzoai	🤖 AI Client Library	pip install hanzoai
	- 100+ LLM providers (OpenAI-compatible interface)
	- Unified gateway and proxy
	- Cost tracking and rate limiting
hanzo-mcp	🛠️ MCP Server & 70+ Tools	pip install hanzo-mcp
	- Complete development environment
	- Interactive notebooks, REPL, debugging
	- Multi-agent workflows and orchestration
hanzo-network	🌐 Decentralized Compute	pip install hanzo-network
	- Distributed AI workload management
	- Peer-to-peer agent networks

Extended Ecosystem

Package	Description
hanzo-agents	Multi-agent workflows and orchestration
hanzo-memory	Vector memory and knowledge management
hanzo-repl	Interactive AI-powered REPL
hanzo-aci	AI-powered development tools

🏗️ Architecture

Local AI Orchestration (Hanzo Zen)

Cost-effective orchestration model that runs on laptops, mobile, or edge devices:

• 90% cost reduction vs pure API calls
• Mixture of Experts (MoE) architecture
• Local orchestration loop with intelligent routing
• API-based LLMs only for complex tasks
• Full MCP tool support and recursive agent calls

Agent Networks

Two deployment architectures:

1. Traditional Swarm: Hierarchical with main orchestrator
2. Peer Networks: All agents equal, connected via MCP
   • Every agent exposes MCP tools to others
   • Recursive calling (chat_with_*, delegate_to_*)
   • Local orchestration with Hanzo Zen

📚 Documentation

The REST API documentation can be found on docs.hanzo.ai. The full API of this library can be found in api.md.

💡 Installation Variants

# Complete ecosystem
pip install hanzo[all]

# Just the CLI and network tools
pip install hanzo

# AI client library (like openai package)  
pip install hanzoai

# MCP development environment
pip install hanzo-mcp

# Specific combinations
pip install hanzo[ai,mcp]    # CLI + AI client + MCP tools
pip install hanzo[agents]    # CLI + agent runtime

🎯 Usage Examples

Command Line Interface

# Start local AI network (90% cost savings)
hanzo net

# Interactive AI chat
hanzo chat

# Launch MCP server for Claude Desktop
hanzo mcp

# Run agent workflows  
hanzo agent --config my-team.yaml

# Check network status
hanzo status

Python SDK - AI Client (hanzoai)

# Use like OpenAI but with 100+ providers
from hanzoai import Hanzo

client = Hanzo(api_key="your-key")
response = client.chat.completions.create(
    model="gpt-4",  # or claude-3-5-sonnet, llama-3, etc.
    messages=[{"role": "user", "content": "Hello!"}]
)

Unified SDK - Everything Together

import hanzo

# Use AI client  
client = hanzo.Client(api_key="your-key")

# Start agent network
network = hanzo.Network()
network.start_agents(["coder", "reviewer", "tester"])

# Use MCP tools
from hanzo.mcp import search, edit, critic
search("authentication bug")
edit("fix auth issue")
critic("review this change")

MCP Development Environment

# Install to Claude Desktop
uvx hanzo-mcp

# Or run standalone  
hanzo-mcp --tools search,edit,notebook,agent

The full API reference can be found in api.md.

While you can provide an api_key keyword argument, we recommend using python-dotenv to add HANZO_API_KEY="sk-xxxxxxx to your .env file so that your API Key is not stored in source control.

Async usage

Simply import AsyncHanzo instead of Hanzo and use await with each API call:

import os
import asyncio
from hanzoai import AsyncHanzo

client = AsyncHanzo(
    api_key=os.environ.get("HANZO_API_KEY"),  # This is the default and can be omitted
)


async def main() -> None:
    response = await client.get_home()


asyncio.run(main())

Functionality between the synchronous and asynchronous clients is otherwise identical.

Using types

Nested request parameters are TypedDicts. Responses are Pydantic models which also provide helper methods for things like:

• Serializing back into JSON, model.to_json()
• Converting to a dictionary, model.to_dict()

Typed requests and responses provide autocomplete and documentation within your editor. If you would like to see type errors in VS Code to help catch bugs earlier, set python.analysis.typeCheckingMode to basic.

Nested params

Nested parameters are dictionaries, typed using TypedDict, for example:

from hanzoai import Hanzo

client = Hanzo()

model = client.model.create(
    hanzo_params={
        "model": "model",
        "api_base": "api_base",
        "api_key": "api_key",
        "api_version": "api_version",
        "aws_access_key_id": "aws_access_key_id",
        "aws_region_name": "aws_region_name",
        "aws_secret_access_key": "aws_secret_access_key",
        "budget_duration": "budget_duration",
        "configurable_clientside_auth_params": ["string"],
        "custom_llm_provider": "custom_llm_provider",
        "input_cost_per_second": 0,
        "input_cost_per_token": 0,
        "hanzo_trace_id": "hanzo_trace_id",
        "max_budget": 0,
        "max_file_size_mb": 0,
        "max_retries": 0,
        "merge_reasoning_content_in_choices": True,
        "model_info": {},
        "organization": "organization",
        "output_cost_per_second": 0,
        "output_cost_per_token": 0,
        "region_name": "region_name",
        "rpm": 0,
        "stream_timeout": 0,
        "timeout": 0,
        "tpm": 0,
        "use_in_pass_through": True,
        "vertex_credentials": "string",
        "vertex_location": "vertex_location",
        "vertex_project": "vertex_project",
        "watsonx_region_name": "watsonx_region_name",
    },
    model_info={"id": "id"},
    model_name="model_name",
)
print(model.hanzo_params)

File uploads

Request parameters that correspond to file uploads can be passed as bytes, a PathLike instance or a tuple of (filename, contents, media type).

from pathlib import Path
from hanzoai import Hanzo

client = Hanzo()

client.audio.transcriptions.create(
    file=Path("/path/to/file"),
)

The async client uses the exact same interface. If you pass a PathLike instance, the file contents will be read asynchronously automatically.

Handling errors

When the library is unable to connect to the API (for example, due to network connection problems or a timeout), a subclass of hanzoai.APIConnectionError is raised.

When the API returns a non-success status code (that is, 4xx or 5xx response), a subclass of hanzoai.APIStatusError is raised, containing status_code and response properties.

All errors inherit from hanzoai.APIError.

import hanzoai
from hanzoai import Hanzo

client = Hanzo()

try:
    client.get_home()
except hanzoai.APIConnectionError as e:
    print("The server could not be reached")
    print(e.__cause__)  # an underlying Exception, likely raised within httpx.
except hanzoai.RateLimitError as e:
    print("A 429 status code was received; we should back off a bit.")
except hanzoai.APIStatusError as e:
    print("Another non-200-range status code was received")
    print(e.status_code)
    print(e.response)

Error codes are as follows:

Status Code	Error Type
400	BadRequestError
401	AuthenticationError
403	PermissionDeniedError
404	NotFoundError
422	UnprocessableEntityError
429	RateLimitError
>=500	InternalServerError
N/A	APIConnectionError

Retries

Certain errors are automatically retried 2 times by default, with a short exponential backoff. Connection errors (for example, due to a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, and >=500 Internal errors are all retried by default.

You can use the max_retries option to configure or disable retry settings:

from hanzoai import Hanzo

# Configure the default for all requests:
client = Hanzo(
    # default is 2
    max_retries=0,
)

# Or, configure per-request:
client.with_options(max_retries=5).get_home()

Timeouts

By default requests time out after 1 minute. You can configure this with a timeout option, which accepts a float or an httpx.Timeout object:

from hanzoai import Hanzo

# Configure the default for all requests:
client = Hanzo(
    # 20 seconds (default is 1 minute)
    timeout=20.0,
)

# More granular control:
client = Hanzo(
    timeout=httpx.Timeout(60.0, read=5.0, write=10.0, connect=2.0),
)

# Override per-request:
client.with_options(timeout=5.0).get_home()

On timeout, an APITimeoutError is thrown.

Note that requests that time out are retried twice by default.

Advanced

Logging

We use the standard library logging module.

You can enable logging by setting the environment variable HANZO_LOG to info.

$ export HANZO_LOG=info

Or to debug for more verbose logging.

How to tell whether None means null or missing

In an API response, a field may be explicitly null, or missing entirely; in either case, its value is None in this library. You can differentiate the two cases with .model_fields_set:

if response.my_field is None:
  if 'my_field' not in response.model_fields_set:
    print('Got json like {}, without a "my_field" key present at all.')
  else:
    print('Got json like {"my_field": null}.')

Accessing raw response data (e.g. headers)

The "raw" Response object can be accessed by prefixing .with_raw_response. to any HTTP method call, e.g.,

from hanzoai import Hanzo

client = Hanzo()
response = client.with_raw_response.get_home()
print(response.headers.get('X-My-Header'))

client = response.parse()  # get the object that `get_home()` would have returned
print(client)

These methods return an APIResponse object.

The async client returns an AsyncAPIResponse with the same structure, the only difference being awaitable methods for reading the response content.

.with_streaming_response

The above interface eagerly reads the full response body when you make the request, which may not always be what you want.

To stream the response body, use .with_streaming_response instead, which requires a context manager and only reads the response body once you call .read(), .text(), .json(), .iter_bytes(), .iter_text(), .iter_lines() or .parse(). In the async client, these are async methods.

with client.with_streaming_response.get_home() as response:
    print(response.headers.get("X-My-Header"))

    for line in response.iter_lines():
        print(line)

The context manager is required so that the response will reliably be closed.

Making custom/undocumented requests

This library is typed for convenient access to the documented API.

If you need to access undocumented endpoints, params, or response properties, the library can still be used.

Undocumented endpoints

To make requests to undocumented endpoints, you can make requests using client.get, client.post, and other http verbs. Options on the client will be respected (such as retries) when making this request.

import httpx

response = client.post(
    "/foo",
    cast_to=httpx.Response,
    body={"my_param": True},
)

print(response.headers.get("x-foo"))

Undocumented request params

If you want to explicitly send an extra param, you can do so with the extra_query, extra_body, and extra_headers request options.

Undocumented response properties

To access undocumented response properties, you can access the extra fields like response.unknown_prop. You can also get all the extra fields on the Pydantic model as a dict with response.model_extra.

Configuring the HTTP client

You can directly override the httpx client to customize it for your use case, including:

• Support for proxies
• Custom transports
• Additional advanced functionality

import httpx
from hanzoai import Hanzo, DefaultHttpxClient

client = Hanzo(
    # Or use the `HANZO_BASE_URL` env var
    base_url="http://my.test.server.example.com:8083",
    http_client=DefaultHttpxClient(
        proxy="http://my.test.proxy.example.com",
        transport=httpx.HTTPTransport(local_address="0.0.0.0"),
    ),
)

You can also customize the client on a per-request basis by using with_options():

client.with_options(http_client=DefaultHttpxClient(...))

Managing HTTP resources

By default the library closes underlying HTTP connections whenever the client is garbage collected. You can manually close the client using the .close() method if desired, or with a context manager that closes when exiting.

from hanzoai import Hanzo

with Hanzo() as client:
  # make requests here
  ...

# HTTP client is now closed

Versioning

This package generally follows SemVer conventions, though certain backwards-incompatible changes may be released as minor versions:

1. Changes that only affect static types, without breaking runtime behavior.
2. Changes to library internals which are technically public but not intended or documented for external use. (Please open a GitHub issue to let us know if you are relying on such internals.)
3. Changes that we do not expect to impact the vast majority of users in practice.

We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience.

We are keen for your feedback; please open an issue with questions, bugs, or suggestions.

Determining the installed version

If you've upgraded to the latest version but aren't seeing any new features you were expecting then your python environment is likely still using an older version.

You can determine the version that is being used at runtime with:

import hanzoai
print(hanzoai.__version__)

Requirements

Python 3.8 or higher.

Contributing

See the contributing documentation.