From e7c971a7bbb82d4d62d7db914e805a45beb03e11 Mon Sep 17 00:00:00 2001
From: Dmytro Skarzhynets <dskarzhynets@thingsboard.io>
Date: Wed, 9 Jul 2025 20:40:14 +0300
Subject: [PATCH] AI rule node: add node description and details

---
 .../thingsboard/rule/engine/ai/TbAiNode.java    | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)

diff --git a/rule-engine/rule-engine-components/src/main/java/org/thingsboard/rule/engine/ai/TbAiNode.java b/rule-engine/rule-engine-components/src/main/java/org/thingsboard/rule/engine/ai/TbAiNode.java
index 93c2cafaae..6c4815308a 100644
--- a/rule-engine/rule-engine-components/src/main/java/org/thingsboard/rule/engine/ai/TbAiNode.java
+++ b/rule-engine/rule-engine-components/src/main/java/org/thingsboard/rule/engine/ai/TbAiNode.java
@@ -52,9 +52,20 @@ import static org.thingsboard.server.dao.service.ConstraintValidator.validateFie
 
 @RuleNode(
         type = ComponentType.EXTERNAL,
-        name = "AI",
-        nodeDescription = "Interact with AI",
-        nodeDetails = "This node makes requests to AI based on a prompt and a input message and returns a response in a form of output message",
+        name = "AI request",
+        nodeDescription = "Sends a request to an AI model using system and user prompts. Supports JSON mode.",
+        nodeDetails = """
+                Interact with large language models (LLMs) by sending dynamic requests from your rule chain.
+                You can select a specific <strong>AI model</strong> and define its behavior using a <strong>system prompt</strong> (optional context or role) and a <strong>user prompt</strong> (the main task).
+                Both prompts can be populated with data and metadata from the incoming message using patterns.
+                For example, the <code>$[*]</code> and <code>${*}</code> patterns allow you to access the all message body and all metadata, respectively.
+                <br><br>
+                After sending the request, the node waits for a response within a configured <strong>timeout</strong>.
+                You can specify the desired <strong>response format</strong> as <strong>Text</strong>, <strong>JSON</strong>, or provide a specific <strong>JSON Schema</strong> to structure the output.
+                The AI-generated content is forwarded as the body of the outgoing message; the originator, message type, and metadata from the incoming message remain unchanged.
+                <br><br>
+                Output connections: <code>Success</code>, <code>Failure</code>.
+                """,
         configClazz = TbAiNodeConfiguration.class,
         ruleChainTypes = RuleChainType.CORE
 )