sema4ai-action-server
Sema4.ai Action Server is a Python framework designed to provide your Python functions to AI Agents. It works both as an MCP Server (hosting tools, resources and prompts) and also provides an OpenAPI compatible API.
A tool or action in this case is defined as a Python function (which has inputs/outputs defined), which is served by the Sema4.ai Action Server.
The Sema4.ai Action Server automatically provides a /mcp endpoint for connecting MCP clients and also generates an OpenAPI spec for your Python code, enabling different AI/LLM Agents to understand and call your Action. It also manages the Action lifecycle and provides full traceability of what happened during any tool or action call (open the /runs endpoint in a browser to see not only inputs and output, but also a full log.html with internal details, such as variables and function calls that your Python function executed).
1. Install Action Server
Action Server is available as a stand-alone fully signed executable and via pip install sema4ai-action-server.
We recommend the executable to prevent confusion in case you have multiple/crowded Python environments, etc.
For macOS
# Install Sema4.ai Action Server
brew update
brew install sema4ai/tools/action-server
For Windows
# Download Sema4.ai Action Server
curl -o action-server.exe https://cdn.sema4.ai/action-server/releases/latest/windows64/action-server.exe
# Add to PATH or move to a folder that is in PATH
setx PATH=%PATH%;%CD%
For Linux
# Download Sema4.ai Action Server
curl -o action-server https://cdn.sema4.ai/action-server/releases/latest/linux64/action-server
chmod a+x action-server
# Add to PATH or move to a folder that is in PATH
sudo mv action-server /usr/local/bin/
2. Run your first MCP tool
# Bootstrap a new project using this template.
# You'll be prompted for the name of the project (directory):
action-server new
# Start Action Server
cd my-project
action-server start
👉 You should now have an Action Server running locally at: http://localhost:8080, so open that in your browser and the web UI will guide you further.
What do you need in your Action Package
An Action Package is currently defined as a local folder that contains at least one Python file containing an action entry point (a Python function marked with @action -decorator from sema4ai.actions).
The package.yaml file is required for specifying the Python environment and dependencies for your Action (RCC will be used to automatically bootstrap it and keep it updated given the package.yaml contents).
Note: the
package.yamlis optional if the action server is not being used as a standalone (i.e.: if it was pip-installed it can use the same python environment where it's installed).
Bootstrapping a new Action
Start new projects with:
action-server new
Note: the action-server executable should be automatically added to your python installation after pip install sema4ai-action-server, but if for some reason it wasn't pip-installed, it's also possible to use python -m sema4ai.action_server instead of action-server.
After creating the project, it's possible to serve the actions under the current directory with:
action-server start
For example: When running action-server start, the action server will scan for existing actions under the current directory, and it'll start serving those.
After it's started, it's possible to access the following URLs:
/index.html: UI for the Action Server./openapi.json: Provides the openapi spec for the action server./docs: Provides access to the APIs available in the server and a UI to test it.
Documentation
Explore our docs for extensive documentation.
Changelog
A list of releases and corresponding changes can be found in the changelog.
1 maintainer
Metadata
1 maintainer
Release files for sema4ai-action-server 3.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| sema4ai_action_server-3.2.0.tar.gz | 474.2 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| sema4ai_action_server-3.2.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 711.1 kB
Release files / sema4ai_action_server-3.2.0.tar.gz
| Download URL | sema4ai_action_server-3.2.0.tar.gz |
|---|---|
| Size | 474.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
894c7a4f92d777cbb158ff5685c37544b491b1fa96e9cd266672a55eac9aab75
|
|
BLAKE2b-256 checksum How to use checksums |
31c616e74abdc79f496febd57263308c9487bf30ded67781b2b48e16a8203cce
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/2.1.4 CPython/3.12.13 Linux/6.8.0-1044-azure
|
Release files / sema4ai_action_server-3.2.0-py3-none-any.whl
| Download URL | sema4ai_action_server-3.2.0-py3-none-any.whl |
|---|---|
| Size | 236.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
1a7ae5ca2ad1bbb8d97f98a69776d817736e3c62e791a6b8f2b45c8bf0d8b15c
|
|
BLAKE2b-256 checksum How to use checksums |
d85f53649fc741a66bbe69429e2de19bb78233c54cd68986a6536f37c5686a13
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
poetry/2.1.4 CPython/3.12.13 Linux/6.8.0-1044-azure
|
