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
- Agent Actions
- API Key Actions
- Chat Actions
- Asks Question in a Given Chat Session. If Stream Is Enabled, the Server Sends Stream of Delta Messages. The Stream Is te...
- Changes the Vote Value of a Chat Message
- Creates Chat Session
- Creates Job to Delete Chat Sessions
- Deletes Agent Server Directories
- Deletes Agent Server Files
- Deletes Collection
- Deletes Specific Chat Messages
- Fetches Chat Message and Metadata for Messages in a Chat Session
- Fetches Chat Message Meta Information
- Fetches Metadata for References of a Chat Message
- Finds a Chat Session by ID
- Gets Stats of a Agent Server Directory
- List Chat Sessions
- List Suggested Questions for a Given Chat Session
- Lists Agent Server Files
- Lists Stats of Agent Server Directories
- Lists User's Questions and Answers that Have a Feedback
- Removes a Collection Reference From the Chat Session
- Removes a Prompt Template Reference From the Chat Session
- Update Feedback for a Specific Answer to a Question
- Updates a Collection Reference of a Chat Session
- Updates a Prompt Template Reference of a Chat Session
- Updates the Name of a Chat Session
- Collection Actions
- Archives a Collection Along with Its Associated Data
- Counts a Number of Chat Sessions with the Collection
- Counts a Number of Documents in the Collection
- Create a Collection
- Creates a Job to Delete Collection Thumbnail
- Creates a Job to Delete Collections
- Creates a Job to Update Collection Thumbnail
- Creates Job to Import Collection to the Collection
- Creates Job to Insert Document to the Collection
- Creates Job to Remove Documents From the Collection
- Deletes Collection
- Deletes Collection Thumbnail
- Fetches All Users' Collection Metadata Sorted by Last Update Time by Default
- Fetches Collection Chat Settings
- Fetches Collection Metadata
- Fetches Collection Settings
- Finds Chunks Related to a Message Using Lexical Search
- Finds Chunks Related to a Message Using Semantic Search
- Get a Collection
- Import an Already Stored Document to an Existing Collection
- List a Collection's Documents
- List Chat Sessions for a Given Collection
- List Collections
- List Suggested Questions for a Given Collection
- Removes a Prompt Template Reference From the Collection
- Removes a Size Limit for a Collection
- Removes an Expiry Date From a Collection
- Removes an Inactivity Interval From the Collection
- Removes Sharing of a Collection to a User
- Removes Sharing of a Collection to All Other Users Except the Original Owner
- Removes the Document From the Collection
- Updates Collection Settings
- Resets the Prompt Settings for a Given Collection
- Restores an Archived Collection to an Active Status
- Returns a List of Access Permissions for a Given Collection
- Returns a List of Group Access Permissions for a Given Collection
- Returns Specific Chunks in a Collection
- Sets a Maximum Limit on the Total Size of Documents (Sum) Added to a Collection
- Shares a Collection to a Group
- Shares a Collection to a User
- Updates a Flag Specifying Whether a Collection Is Private or Public
- Updates a Prompt Template Reference of a Collection
- Updates an Expiry Date of a Collection
- Updates an Inactivity Interval of a Collection
- Updates Attributes of an Existing Collection
- Updates Collection Chat Settings
- Updates Collection Metadata
- Updates Collection Thumbnail
- Configuration Actions
- Deletes Global Configuration Items
- Deletes Role Configuration Items
- Deletes User Configuration Items
- Gets Configurations for a Given Role
- Gets Global Configurations
- Gets User Configurations
- Resets User Configuration Item
- Sets Configuration Item for a Given Role
- Sets Global Configuration Item
- Sets User Configuration Item
- Document Actions
- Assigns a Tag to the Document
- Counts a Number of Chat Sessions with the Document
- Counts a Number of Documents
- Creates Job to Delete Documents
- Creates Job to Process Document
- Deletes a Document
- Deletes Document Summaries
- Fetches Document Guardrails Settings
- Fetches Document Internal Metadata
- Fetches Document Metadata
- Fetches Document Page Layout
- Fetches Document Page Ocr Model
- Fetches Document Summary
- Fetches Document User Source File
- Fetches Recent Document Summaries/extractions/transformations
- Finds a Document by ID
- List Chat Sessions for a Given Document
- List Documents
- Lists Collections for Containing a Given Document
- Removes a Tag From a Document
- Returns All Chunks for a Specific Document
- Updates Attributes of an Existing Document
- Updates Document Metadata
- Document Ingestion Actions
- Adds Files From the AWS S3 Storage Into a Collection
- Adds Files From the Azure Blob Storage Into a Collection
- Adds Files From the Google Cloud Storage Into a Collection
- Adds Files From the Local System Into a Collection
- Adds Plain Text to a Collection
- Converts Files Uploaded in "Agent_only" Ingest Mode to PDF and Parses Them
- Crawls and Ingest a URL Into a Collection
- Creates a Job to Add Files From the AWS S3 Storage Into a Collection
- Creates a Job to Add Files From the Azure Blob Storage Into a Collection
- Creates a Job to Add Files From the Google Cloud Storage Into a Collection
- Creates a Job to Add Files From the Local System Into a Collection
- Creates a Job to Add Plain Text to a Collection
- Creates a Job to Crawl and Ingest a URL Into a Collection
- Creates a Job to Ingest Uploaded Document
- Creates a Job to Parse Files Uploaded in "Agent_only" Ingest Mode
- Ingest Uploaded Document
- Uploads File to H2OGPTe Instance
- Job Actions
- Model Actions
- Creates a Topic Model on the Collection
- Creates Job for Creation of a Topic Model
- Encode Texts for Semantic Searching
- Extract Information From One or More Contexts Using an LLM
- Helper to Get Reasonable (Easy to Use) Defaults for Guardrails/PII Settings
- Returns Performance Statistics Grouped by Models
- Returns Usage Statistics for All Models
- Returns Usage Statistics Grouped by Models
- Returns Usage Statistics Grouped by Models and Users
- Returns Usage Statistics Grouped by Users
- Runs a Self-Test for a Given Model
- Send a Message and Get a Response From an LLM
- Summarize One or More Contexts Using an LLM
- Permission Actions
- Deletes Roles for Given Unique Identifiers
- Assigns Permission to a Given Role
- Assigns Roles to a Given Group
- Assigns Roles to a Given User
- Associates a User with a Document They Have Permission On
- Checks if Collection Permission Is Granted for a Given User
- Checks if Permission Is Granted for a Given User
- Creates a Role
- Creates a User Group
- Deletes Groups for Given Group Names
- Deletes Groups for Given Unique Identifiers
- Deletes Roles for Given Role Names
- Finds Role for a Given Unique Identifier
- Finds Roles Associated with a Given Group
- Finds Roles Associated with a Given User
- Finds User for a Given Unique Identifier
- Lists Permissions of a Given Group
- Lists Permissions of a Given Role
- Lists Permissions of a Given User
- Removes Permission From a Given Role
- Removes Roles From a Given Group
- Removes Roles From a Given User
- Resets Group Roles
- Resets User Roles
- Returns a List of All Registered Users for the System
- Returns System Permissions
- Sets a New Set of Permissions for a Given Role
- Sets Priority for a Given Role
- Prompt Template Actions
- Creates a New Prompt Template
- Deletes a Prompt Template
- Finds a Prompt Template by ID
- List Prompt Templates
- Lists Prompt Templates, Including Hidden Default Templates
- Removes Access to a Prompt Template for a Group
- Removes Sharing of a Prompt Template to a User
- Removes Sharing of a Prompt Template to All Other Users Except the Original Owner
- Resets and Shares a Prompt Template to a New List of Groups
- Resets and Shares a Prompt Template to a New List of Users
- Returns a List of Access Permissions for a Given Prompt Template
- Returns a List of Group Access Permissions for a Given Prompt Template
- Shares a Prompt Template to a User
- Shares a Prompt Template with a Group
- Updates a Flag Specifying Whether a Default Prompt Template Is Visible or Hidden to Users
- Updates Attributes of a Given Prompt Template
- Tag Actions
Overview
This node allows users to ask a question within an existing chat session in the h2oGPT environment. It sends a query or instruction to a large language model (LLM) associated with the specified chat session and retrieves the response. If streaming is enabled, the server sends a stream of incremental delta messages, allowing real-time updates as the answer is generated.
Common scenarios for this node include:
- Integrating conversational AI into workflows where ongoing chat sessions are maintained.
- Querying knowledge bases or documents through a chat interface powered by LLMs.
- Using streaming responses to display partial answers immediately for better user experience.
Practical example:
- A user wants to ask "Who is the strongest hero?" in a specific chat session that has context from previous interactions. The node sends this message to the LLM and returns the answer, optionally streaming it if enabled.
Properties
| Name | Meaning |
|---|---|
| Session ID | ID of the chat session where the question will be asked. This identifies the conversation context. |
| Message | The query or instruction text sent to the LLM. For example, a question like "Who is the strongest hero?". |
| Additional Options | A collection of optional parameters to customize the request: |
| - Image Batch Final Prompt | A prompt applied to each image batch for vision-capable models. |
| - Image Batch Image Prompt | A prompt to summarize all answers per image batch for vision models. |
| - Include Chat History | Whether to include previous questions and answers from the current chat session in the context for the new query. |
| - Llm | The name of the LLM to use for answering. Use "auto" for automatic model selection. |
| - Llm Args | JSON map of arguments for the LLM, e.g., temperature to control randomness. |
| - Pre Prompt Query | Text prepended before contextual document chunks to guide the LLM. |
| - Prompt Query | Text appended at the beginning of the user's message to customize prompt behavior. |
| - Rag Config | JSON map controlling retrieval-augmented generation (RAG) settings, such as rag_type. |
| - Self Reflection Config | JSON map with self-reflection settings for the LLM, e.g., specifying reflection prompts or models. |
| - Stream | Boolean flag to enable streaming of incremental response messages from the server. |
| - System Prompt | Text sent as a system prompt to models supporting it, providing overall context on how to respond. Use 'auto' for default behavior. |
| - Tags | List of tags used to pull context for RAG. |
Output
The node outputs the full response from the LLM in the json field of the output data. This includes the answer generated by the model based on the input message and chat session context.
If streaming is enabled, the output represents a stream of delta messages incrementally updating the answer as it is generated by the server.
The node does not explicitly output binary data.
Dependencies
- Requires connection to an h2oGPT API endpoint, configured via credentials including the base URL and an API key credential.
- The node uses HTTP POST requests to the
/chats/{session_id}/completionsendpoint of the h2oGPT API. - Proper authentication credentials must be set up in n8n to access the h2oGPT service.
Troubleshooting
- Invalid Session ID: If the provided session ID does not exist or is invalid, the API may return an error. Verify the session ID is correct and active.
- Missing Required Parameters: Both
session_idandmessageare required. Omitting these will cause errors. - Streaming Issues: When enabling streaming, ensure your environment supports handling streamed responses; otherwise, incomplete or failed responses may occur.
- API Authentication Errors: Ensure the API key credential is valid and has permissions to access the chat session.
- Timeouts: Large or complex queries might take longer; consider adjusting timeout settings if available or simplifying the query.