ambi-python 0.1.1

A flexible, multi-backend, customizable AI agent framework, entirely based on Rust.

Python Bindings for Ambi

A Python native extension for the Ambi AI Agent framework.

Installation

From PyPI (recommended)

pip install ambi-python

Build from source

Requires maturin and a Rust toolchain.

# Install maturin
pip install maturin

# Build and install locally
cd bindings/python
maturin develop --release

Then import as:

from ambi import Agent, AgentState, Pipeline, LLMEngineConfig

Build & Publish

cd bindings/python

# Build wheel
maturin build --release

# Publish to PyPI
maturin publish --username __token__ --password pypi-xxxxx

The built .whl files will be in target/wheels/.

Examples

Run any example from the bindings/python directory:

cd bindings/python

# Basic cloud-based chat
OPENAI_API_KEY=sk-... python examples/chat-cloud.py

# Streaming output
OPENAI_API_KEY=sk-... python examples/chat-stream.py

# Tool calling
OPENAI_API_KEY=sk-... python examples/custom-tool.py

# Custom chat template
OPENAI_API_KEY=sk-... python examples/custom-template.py

# Custom Python LLM engine
python examples/custom-engine.py

# Memory eviction
OPENAI_API_KEY=sk-... python examples/memory-eviction.py

API Reference

Python	JS equivalent	Description
Agent.make(config)	Agent.make(config)	Create an async agent
agent.preamble("...")	agent.preamble("...")	Set system prompt
agent.template("chatml")	agent.template(Chatml)	Set template type
agent.custom_template(...)	agent.customTemplate(...)	Custom template
agent.add_tool(...)	agent.tool(...)	Register a tool
agent.with_standard_formatting()	agent.withStandardFormatting()	Enable standard formatting
agent.with_eviction_strategy(...)	agent.withEvictionStrategy(...)	Set eviction strategy
agent.count_tokens(...)	agent.countTokens(...)	Count tokens
AgentState("id")	new AgentState("id")	Create session state
LLMEngineConfig.openai(...)	LLMEngineConfig.openai(...)	OpenAI engine
LLMEngineConfig.custom(...)	LLMEngineConfig.custom(...)	Custom engine
Pipeline.chat_runner(n)	Pipeline.chatRunner(n)	Chat runner
Pipeline.custom(...)	Pipeline.custom(...)	Custom pipeline
runner.chat(...)	runner.chat(...)	Sync chat
runner.chat_stream(...)	runner.chatStream(...)	Streaming chat
resolve_request(id, result)	resolveRequest(id, result)	Resolve async callback
resolve_pipeline_request(...)	resolvePipelineRequest(...)	Resolve pipeline callback