h2oGPTe icon

h2oGPTe

h2oGPTe is an AI-powered search assistant for your internal teams to answer questions gleaned from large volumes of documents, websites and workplace content.

Actions198

Overview

This node operation creates a new prompt template within the system. Prompt templates are reusable sets of instructions or prompts that can be associated with collections to guide AI models in generating responses, summaries, or other outputs based on documents and user queries. This is particularly useful for customizing how AI interacts with your data, enabling consistent and tailored AI behavior across different collections or projects.

Practical examples include:

  • Creating a prompt template that defines how to summarize documents in a specific style.
  • Defining sample questions automatically generated for newly imported documents.
  • Setting up system prompts or reflection prompts to improve AI response quality.

Properties

Name Meaning
Name The name of the prompt template (required).
Additional Options A collection of optional settings to further customize the prompt template:
- Auto Gen Description Prompt A prompt used to automatically create a description of the collection.
- Auto Gen Document Sample Questions Prompt A prompt to generate sample questions for a freshly imported document if enabled.
- Auto Gen Document Summary Pre Prompt Summary A pre-prompt summary text for summarizing a freshly imported document if enabled.
- Auto Gen Document Summary Prompt Summary A prompt summary text for summarizing a freshly imported document if enabled.
- Default Sample Questions Default sample questions used if no auto-generated questions are available.
- Description A textual description of the prompt template.
- Hyde No Rag Llm Prompt Extension An extension prompt for LLM when not using retrieval-augmented generation (RAG).
- Image Batch Final Prompt A prompt applied to each image batch for vision models.
- Image Batch Image Prompt A prompt for each image batch for vision models.
- Lang Language code specifying the language context of the prompt template.
- Pre Prompt Query Text prepended before contextual document chunks.
- Pre Prompt Summary A prompt placed before each large piece of text to summarize.
- Prompt Query Text appended at the beginning of the user's message.
- Prompt Reflection A self-reflection template requiring two occurrences of %s for full previous prompt inclusion.
- Prompt Summary A prompt placed after each large piece of text to summarize.
- System Prompt A system-level prompt providing overall context to the AI model.
- System Prompt Reflection A system prompt designed for self-reflection by the AI model.

Output

The output of this operation is the JSON response from the API representing the newly created prompt template. This typically includes all properties of the prompt template such as its unique identifier, name, description, and any additional options set during creation.

If the API supports it, the output may also include metadata about the prompt template's creation status or timestamps.

No binary data output is expected from this operation.

Dependencies

  • Requires an API key credential for authentication to the backend service.
  • The node sends a POST request to the /prompt_templates endpoint of the configured API base URL.
  • Proper network connectivity and valid credentials are necessary for successful execution.

Troubleshooting

  • Missing Required Fields: Ensure the "Name" property is provided; it is required.
  • Authentication Errors: Verify that the API key credential is correctly configured and has sufficient permissions.
  • API Endpoint Issues: Confirm the base URL and endpoint /prompt_templates are accessible and correct.
  • Invalid Option Values: Check that optional fields, if provided, conform to expected formats (e.g., strings where strings are expected).
  • Timeouts or Network Failures: Increase timeout settings or check network connectivity if requests fail due to timeouts.

Links and References

Discussion