RenderActionDefinition

RenderActionDefinitions are templates that describe what StoryEngine should output when a goal completes. GoalDefinitions can have zero or more RenderActionDefinitions, depending on their needs. This lets StoryEngine output multiple messages in different formats as the narrative progresses.

Engines

StoryEngine supports three render engines for RenderActions.

Template

The template engine evaluates the configured Prompt and then directly emits the resulting output. This engine is useful to send specific, structured messages back to the game that do not require any input from an AI. For example, a RenderActionDefinition could be created to send a JSON message to inform the game a specific goal has completed.

Anthropic

The Anthropic engine uses Anthropic with the System Prompt’s configured model. Both the configures System Prompt and Prompt are evaluated, then sent to Anthropic for processing. The response is then send back to the game as a message for use.

OpenAI

The OpenAI engine uses OpenAI with the System Prompt’s configured gpt version. Both the configured System Prompt and Prompt are evaluated, then sent to OpenAI for processing. The response is then sent back to the game as a message for use.

Prompts

Prompts are used to generate output that is sent back to the game.

System Prompt

The system prompt is only used by the OpenAI and Anthropic engines.

See SystemPrompt for details.

Prompt

The prompt is evaluated with the current StoryEngine state to generate output.

See Prompt for details.

Relevant History Prompt

The relevant history prompt is used to generate embeddings that can provide additional context when processing the prompt. This prompt is optional.

Additional Properties

Include In Summarization

This flag lets you enable or disable summarization for the RenderAction.

StoryEngine internally summarizes RenderActions to maintain a running context for each actor. These summarizations are then available for use within prompts to provide additional context.

Not all RenderActions are suitable for summarization. For example, you may not want to summarize the goal completed message from the example above.

Tag

Tags are game defined values that are included in messages sent from StoryEngine. This helps games route the message and know how to properly interpret the message payload.