autonomy-dev 0.2.160

A collection of tooling to enable open source development of autonomy tools

Autonomy Dev

Tooling to speed up open-autonomy development.

For detailed instructions please see the Docs.

TLDR

# Install adev
pip install -U "autonomy-dev[all]"
# Make a new agent
adev create author/cool_agent
# Run the new agent
adev run dev author/cool_agent

Requirements

• Python >= 3.10,<3.12
• Poetry >= 1.8.3
• Docker Desktop

Features

• Scaffolding of new repositories, agent, protocols, contracts, connections, skills.
• Linting and formatting tools
• Dependency management tools
• Test suite scaffolding

Creating New Github Projects

We can make an autonomy repo

adev repo scaffold fun_new_hack
cd fun_new_hack

Creating a new Agent

Once we have a new project, we can build new agents from templates There are a number of templates available and the --help flag will display the available options.

adev create author/cool_agent

By default, we provide a simple server with ping pong via websockets available at localhost:5555

# run the agent and verify the endpoint
adev run author/cool_agent

Creating an Agent Service

Running agents in production requires a service to manage the agent lifecycle. We can convert an agent to a service with the following command.

adev convert agent-to-service author/agent_name author/service_name

Scaffolding of Components

Contracts

We provide tools to scaffold smart contract components from existing deployed contracts. The scaffolding process includes:

• Complete open-aea contract component
• Contract class with auto-generated methods
• Events, Read and Write methods extracted from the contract ABI.
• Type hints and documentation

# Scaffold USDC contract from Base
adev scaffold contract author/usdc \
    --address 0x833589fcd6edb6e08f4c7c32d4f71b54bda02913 \
    --network base

Protocols

Protocols components can be fully scaffolded from a yaml file. The scaffolding process includes:

• Protocol class with auto-generated methods
• Linted and formatted code
• Type hints and documentation
• Test Suite
• Pydantic models for custom types.

adev scaffold protocol auto_dev/data/protocols/examples/balances.yaml

Dependency Management

For projects created with adev, updates to both:

• autonomy packages
• upstream python packages

are automated using as so;

adev deps verify

Release

checkout main
git pull
adev release

Test Coverage

<!-- Pytest Coverage Comment:Begin -->
Name                             Stmts   Miss  Cover   Missing
--------------------------------------------------------------
auto_dev/__init__.py                 0      0   100%
auto_dev/base.py                    60     19    68%   66-88
auto_dev/check_dependencies.py     236    236     0%   28-452
auto_dev/cli.py                      4      1    75%   9
auto_dev/cli_executor.py            68     36    47%   33-61, 79, 83, 87-89, 92-94, 99-105
auto_dev/constants.py               25      0   100%
auto_dev/enums.py                   36      0   100%
auto_dev/exceptions.py               5      0   100%
auto_dev/fmt.py                     59     43    27%   16-17, 21-22, 27-45, 50, 60-61, 66-80, 85-97, 102-112
auto_dev/lint.py                     7      3    57%   13-27
auto_dev/local_fork.py              52     32    38%   32-33, 37-54, 58-95
auto_dev/test.py                    16     13    19%   4, 16-39
auto_dev/utils.py                  251    153    39%   76-77, 81, 94-101, 106-151, 167, 180-185, 204-228, 233, 240-242, 247, 252, 257-269, 276-281, 290-293, 298-318, 323-337, 342-348, 370-372, 381, 388-416
--------------------------------------------------------------
TOTAL                              819    536    35%
<!-- Pytest Coverage Comment:End -->

Miscillaneous

There are a number of useful command tools available.

• Dev Tooling: A). linting adev lint B). formatting adev fmt C). dependency management adev deps update

• Scaffolding: Tooling to auto generate repositories and components.

Documentation

Running Docs Locally

To run and preview the documentation locally:

# Install mkdocs and required dependencies
pip install mkdocs-material mkdocstrings[python] mkdocs-include-markdown-plugin mkdocs-mermaid2-plugin

# Serve the documentation (available at http://127.0.0.1:8000)
mkdocs serve

This will start a local server and automatically reload when you make changes to the documentation.