Skip to main content

plugin-llm

plugin-llm provides a Turing-workflow actor that sends prompts to LLM providers (Claude, OpenAI, etc.) and returns the response as a workflow action result.

Maven Dependency​

<dependency>
    <groupId>com.scivicslab.turingworkflow.plugins</groupId>
    <artifactId>plugin-llm</artifactId>
    <version>1.0.0</version>
</dependency>

Actions​

See the Javadoc for full argument specifications.

ActionDescription
setProviderSet the LLM provider (claude, openai, …)
setModelSet the model name (e.g. claude-opus-4-6)
setApiKeySet the API key (or read from environment variable)
promptSend a prompt string and return the LLM response

Usage in YAML​

name: llm-example
steps:
  # Load and instantiate the LLM actor
  - states: ["0", "1"]
    actions:
      - actor: loader
        method: loadJar
        arguments: "com.scivicslab.turingworkflow.plugins:plugin-llm:1.0.0"

  - states: ["1", "2"]
    actions:
      - actor: loader
        method: createChild
        arguments: ["ROOT", "llm", "com.scivicslab.turingworkflow.plugins.llm.LlmActor"]

  # Configure the actor
  - states: ["2", "3"]
    actions:
      - actor: llm
        method: setProvider
        arguments: "claude"
      - actor: llm
        method: setModel
        arguments: "claude-opus-4-6"

  # Send a prompt
  - states: ["3", "end"]
    actions:
      - actor: llm
        method: prompt
        arguments: "Summarize the actor model in one sentence."

MCP Gateway Integration​

plugin-llm works with quarkus-mcp-gateway to route LLM calls through an MCP server. When configured with an MCP gateway URL, the actor forwards prompts to the gateway instead of calling the LLM provider directly, enabling tool use and multi-step agent workflows within a single YAML workflow execution.

References​