updated

Author: Nursid

docs/api/apiaccess/index.md

@@ -20,6 +20,7 @@ cbapicategory:
     - name: CodeUtils
       link: /docs/api/apiaccess/codeutils
       description: This is a module that provides various utilities for parsing and manipulating code.
+
     - name: Crawler
       link: /docs/api/apiaccess/crawler
       description: This is a module that crawls the web and returns the crawled data.

docs/api/codeboltutils/_category_.json

@@ -0,0 +1,11 @@
+{
+    "label": "CodeBoltUtils",
+    "position": 2,
+    "collapsible": false,
+    "collapsed": true,
+    "className": "red",
+    "link": {
+      "type": "doc",
+      "id": "api/codeboltutils/index"
+    }
+  }

docs/api/codeboltutils/followupquestionbuilder.md

@@ -0,0 +1,50 @@
+---
+name: FollowUpQuestionBuilder
+cbbaseinfo:
+  description: Creates the next prompt for the AI agent by incorporating previous conversation, tool results, and summarizing long interactions when needed.
+cbparameters:
+  parameters: []
+  returns:
+    signatureTypeName: InferenceParams
+    description: Returns a new inference prompt to continue the agent workflow.
+  typeArgs: []
+data:
+  name: FollowUpQuestionBuilder
+  category: codeboltutils
+  link: followupquestionbuilder.md
+---
+<CBBaseInfo/>
+<CBParameters/>
+
+### Example
+
+```javascript
+
+const codebolt = require('@codebolt/codeboltjs').default;
+
+const { FollowUpPromptBuilder } = require("@codebolt/codeboltjs/utils");
+
+const followUpBuilder = new FollowUpQuestionBuilder(codebolt);
+const nextPrompt = await followUpBuilder
+    .addPreviousConversation(previousPrompt)
+    .addToolResult(toolResults)
+    .checkAndSummarizeConversationIfLong()
+    .buildInferenceParams();
+```
+
+### Methods
+
+The FollowUpQuestionBuilder class provides the following chainable methods:
+
+- **`addPreviousConversation(previousPrompt)`**: Adds the previous conversation context to the new prompt
+- **`addToolResult(toolResults)`**: Incorporates tool execution results into the prompt
+- **`checkAndSummarizeConversationIfLong()`**: Automatically summarizes the conversation if it becomes too long
+- **`buildInferenceParams()`**: Builds and returns the final inference parameters for the next AI interaction
+
+### Usage Notes
+
+- All methods return the FollowUpQuestionBuilder instance, allowing for method chaining
+- The constructor requires a `codebolt` parameter
+- Use this class to maintain conversation context across multiple AI interactions
+- The summarization feature helps manage long conversations by condensing previous interactions
+- The final `buildInferenceParams()` method returns an `InferenceParams` object ready for the next LLM call

docs/api/codeboltutils/index.md

@@ -0,0 +1,17 @@
+---
+cbapicategory:
+  - name: PromptBuilder
+    link: /docs/api/apiaccess/codeboltutils/promptBuilder
+    description: Constructs a smart prompt for the AI agent by combining tools, environment details, system instructions, and task-specific information.
+    
+  - name: LLMOutputHandler
+    link: /docs/api/apiaccess/codeboltutils/llmoutputhandler
+    description: Processes the AI model’s response, manages tool executions, handles user communication, and detects when the task is completed.
+    
+  - name: FollowUpQuestionBuilder
+    link: /docs/api/apiaccess/codeboltutils/followupquestionbuilder
+    description: Creates the next prompt for the AI agent by incorporating previous conversation, tool results, and summarizing long interactions when needed.
+
+---
+# Codebolt Utils
+<CBAPICategory />

docs/api/codeboltutils/llmoutputhandler.md

@@ -0,0 +1,50 @@
+---
+name: LLMOutputHandler
+cbbaseinfo:
+  description: Processes the AI model's response, manages tool executions, handles user communication, and detects when the task is completed.
+cbparameters:
+  parameters: []
+  returns:
+    signatureTypeName: void
+    description: Executes AI logic and tool workflows, and optionally returns completion status.
+  typeArgs: []
+data:
+  name: LLMOutputHandler
+  category: codeboltutils
+  link: llmoutputhandler.md
+---
+<CBBaseInfo/>
+<CBParameters/>
+
+### Example
+
+```javascript
+const codebolt = require('@codebolt/codeboltjs').default;
+
+const { LLMOutputHandler } = require("@codebolt/codeboltjs/utils");
+
+const llmOutput = codebolt.llm.inference(prompt);
+const outputHandler = new LLMOutputHandler(llmOutput, codebolt);
+
+await outputHandler.sendMessageToUser();
+const toolResults = await outputHandler.runTools();
+
+if (outputHandler.isCompleted()) {
+    console.log("Task completed!");
+}
+```
+
+### Methods
+
+The LLMOutputHandler class provides the following methods:
+
+- **`sendMessageToUser()`**: Sends the AI model's response message to the user
+- **`runTools()`**: Executes any tools that were called by the AI model and returns the results
+- **`isCompleted()`**: Checks if the current task has been completed based on the AI's response
+
+### Usage Notes
+
+- The constructor requires `llmOutput` (the AI model's response) and `codebolt` parameters
+- Use `sendMessageToUser()` to communicate AI responses to the user
+- Call `runTools()` to execute any tool calls made by the AI
+- Check `isCompleted()` to determine if the workflow should continue or terminate

docs/api/codeboltutils/promptBuilder.md

@@ -0,0 +1,52 @@
+---
+name: PromptBuilder
+cbbaseinfo:
+  description: Constructs a smart prompt for the AI agent by combining tools, environment details, system instructions, and task-specific information.
+cbparameters:
+  parameters: []
+  returns:
+    signatureTypeName: InferenceParams
+    description: Returns an inference parameter object suitable for running an LLM.
+  typeArgs: []
+data:
+  name: PromptBuilder
+  category: codeboltutils
+  link: promptBuilder.md
+---
+<CBBaseInfo/>
+<CBParameters/>
+
+### Example
+
+```javascript
+
+const codebolt = require('@codebolt/codeboltjs').default;
+
+const { InitialPromptBuilder } = require("@codebolt/codeboltjs/utils");
+
+const promptBuilder = new InitialPromptBuilder(userMessage, codebolt);
+const prompt = await promptBuilder
+    .addMCPTools()
+    .addAgentTools()
+    .addEnvironmentDetails()
+    .addSystemPrompt('agent.yaml', 'test', 'example.md')
+    .addTaskInstruction('task.yaml', 'main_task')
+    .buildInferenceParams();
+```
+
+### Methods
+
+The PromptBuilder class provides the following chainable methods:
+
+- **`addMCPTools()`**: Adds MCP (Model Context Protocol) tools to the prompt
+- **`addAgentTools()`**: Includes agent-specific tools in the prompt
+- **`addEnvironmentDetails()`**: Adds environment context and details
+- **`addSystemPrompt(yamlFile, section, exampleFile)`**: Loads system instructions from YAML configuration
+- **`addTaskInstruction(yamlFile, taskName)`**: Adds specific task instructions
+- **`buildInferenceParams()`**: Builds and returns the final inference parameters
+
+### Usage Notes
+
+- All methods return the PromptBuilder instance, allowing for method chaining
+- The constructor requires `userMessage` and `codebolt` parameters
+- The final `buildInferenceParams()` method returns an `InferenceParams` object ready for LLM execution