The Matchmaking Agent is a module of the Tractus-X Knowledge Agents Reference Implementations.
Up to now the Matchmaking Agent's functionality was integrated into the Agent Plane Container. This container separates the Matchmaking Agent's functionality from the Agent Plane Container. As a result, the RDF Store is disjoined from the public-facing Agent Plane container. It caters for the need of small and medium sized companies who want to directly upload/provide use case data as well as for security-aware companies who want to separate data storage from public API even for meta-data. This container also has the ability to be used by the Simple Graph Exchanger Feature, adding the ability to host business data from the RDF Store over the Matchmaking Agent Communication between the Agent Plane Container and the Matchmaking Agent by a REST API on the docker (internal) network.
The FOSS Matchmaking Agent is built using Jersey (for the REST endpoints) and RDF4J (for SPARQL parsing). It exposes three endpoints which are to be certified/tested in the KA architecture:
- With the '/api/agent' endpoint the EDC tenant can issue queries and execute skills in interaction with local resources and the complete Dataspace (the so-called Matchmaking Agent that is also hit by incoming Agent transfers).
- With the '/api/graph' endpoint the EDC tenant can manage graph content which can be published as assets.
- With the '/agentsource' endpoint communication with the EDC Dataplane is carried out. However, this is a docker internal endpoint and can not be accessed by a user from the internet
The Matchmaking Agent does not perform any logic but just returns pre-rendered results depending on the input parameters.
Architecture is similar to the one shown in Tractus-X Knowledge Agents EDC Extensions (KA-EDC). However the AgentSource Object within the Dataplane accesses the AgentSourceController of the MatchmakingAgent by a REST API.
The conforming agent will be configurable to support the following http/s based security mechanisms:
- Basic Authentication
- Oauth2
mvn -Pwith-docker-image
This will generate
- a jar file in the respective target folder
- a docker image within the local docker repository which can be started with
The standalone jar may be started as follows
java -Dproperty.file.location="dataplane.properties" -cp ../matchmaking-agent-1.14.24-SNAPSHOT.jar org.eclipse.tractusx.agents.conforming.Bootstrap
Make sure that jar file, properties file and dataspace.ttl are in the same directory Then you should be able to reach the /graph endpoint
Afterwards, you should be able to access the local endpoint by directly invoking a query
curl --location --globoff 'localhost:8280/api/agent?query=SELECT%20%3Fsubject%20%3Fpredicate%20%3Fobject%20WHERE%20{%20%3Fsubject%20%3Fpredicate%20%3Fobject.}'
docker run -i -p 8280:8280 -t <imageID>
Afterwards, you should be able to access the local endpoint by directly invoking a query
curl --location --globoff 'localhost:8280/api/agent?query=SELECT%20%3Fsubject%20%3Fpredicate%20%3Fobject%20WHERE%20{%20%3Fsubject%20%3Fpredicate%20%3Fobject.}'
Nevertheless it is intended that the Matchmaking Agent does not run on its own but rather in a combination of a EDC Data Plane and a Control plane as depicted in the diagram above
DockerHub: https://hub.docker.com/r/tractusx/matchmaking-agent
Eclipse Tractus-X product(s) installed within the image:
- GitHub: https://github.com/eclipse-tractusx/knowledge-agents/tree/main/matchmaking
- Project home: https://projects.eclipse.org/projects/automotive.tractusx
- Dockerfile: https://github.com/eclipse-tractusx/knowledge-agents/blob/main/matchmaking/src/main/docker/Dockerfile
- Project license: Apache License, Version 2.0
Used base image
- eclipse-temurin:23-jre-alpine
- Official Eclipse Temurin DockerHub page: https://hub.docker.com/_/eclipse-temurin
- Eclipse Temurin Project: https://projects.eclipse.org/projects/adoptium.temurin
- Additional information about the Eclipse Temurin images: https://github.com/docker-library/repo-info/tree/master/repos/eclipse-temurin
As with all Docker images, these likely also contain other software which may be under other licenses (such as Bash, etc from the base distribution, along with any direct or indirect dependencies of the primary software being contained).
As for any pre-built image usage, it is the image user's responsibility to ensure that any use of this image complies with any relevant licenses for all software contained within.