agntcy-acp 1.5.2

Agent Connect Protocol SDK

Agent Connect Protocol SDK

About The Project

The "Agent Connect Protocol SDK" is an open-source library designed to facilitate the adoption of the Agent Connect Protocol. It offers tools for both client and server implementations, enabling seamless integration and communication between multi-agent systems.

Getting Started

See Getting Started Guide

Documentation

• See ACP SDK Documentation to deep dive into using the client SDK.
• See IoA Documentation for more info on Internet of Agents.

Testing

Run make test in the root of the repo. Tool requirements:

• uv (preferred) or poetry (deprecated): to manage Python dependencies
• make: to store command recipes

Building the package

⚠️ Note: this step is only necessary for maintainers of the project. ⚠️

The agntcy-acp Python package is built on GitHub and published using GitHub actions. The action can be found in the relevant workflows directory in the repo. The project attempts to keep the SDK updated on any ACP or relevant specification changes, but delays can happen.

Prerequisites

This repo uses the following tools to build (or update) the packages:

• jq: to parse OpenAPI JSON
• uv (preferred) or poetry (deprecated): to manage Python dependencies
• make: to store command recipes
• docker: to run the openapi-generator-cli tool
• git: to checkout the source specifications

Generating the SDK clients from the OpenAPI ACP specification

There are two make targets to generate the clients:

• make generate_acp_client
• make generate_acp_async_client

Note that the make targets add a SPDX header and update the package imports to match the files as they should appear in the agntcy_acp/acp_vXX subpackages. Please check the Makefile for questions on how this is done.

To update the agntcy_acp package by copying the relevant files, use: make update_python_subpackage

Updating the client package on a new ACP specification release

For a minor release, follow these steps:

1. Run: ACP_SPEC_RELEASE=<RELEASE_TAG> make update_python_subpackage using the relevant "<RELEASE_TAG>"
2. Check for any irregularities: git diff
3. Run: make test

For a major release, follow these steps:

1. Run: ACP_SPEC_RELEASE=<RELEASE_TAG> make update_python_subpackage using the relevant "<RELEASE_TAG>"
2. Update the version imports if you want to change the default major version in:
   • agntcy_acp/__init__.py
   • agntcy_acp/models/__init__.py
3. Check for any irregularities: git diff
4. Run: make test

Generate API documentation on code updates

Use the make target:

make docs

Publishing

Publishing the package uses a GitHub action triggered by assigning tags to commits of the form v<PACKAGE_VERSION>.

The tag must correspond to the version in the pyproject.toml file, except for dev releases. The tag for a dev release will be v<PACKAGE_VERSION>.devN where the .devN suffix is not part of the package version in the pyproject.toml file. All tags, except dev releases, should be applied to the main branch. Dev releases can be applied anywhere and thus are not guaranteed to be repeatable (e.g., when applied to a PR that is later merged and the branch is deleted).

Not all PEP-440 tags are supported at this time.

The following steps are required to create a release:

1. Push a properly formatted tag (vX.Y.Z[aN][.devN]).
2. Use the specified tag to create a release

The publish action can also be triggered on a branch through the UI. In this case, it will use the package version at the head of the branch.

Roadmap

See the open issues for a list of proposed features and known issues.

Contributing

Contributions are what make the open-source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated. For detailed contributing guidelines, please see CONTRIBUTING.md.

Copyright Notice

Copyright Notice and License

Distributed under Apache 2.0 License. See LICENSE for more information. Copyright AGNTCY Contributors (https://github.com/agntcy)

Acknowledgements

This SDK is developed with the support of the IoA community with the goal of facilitating cross-framework agent interoperability.