Instrument Pendo Track Agent

[novus-by-pendo]

Summary

Add Pendo trackAgent() instrumentation to the LLMConversationView component to track user interactions with the AI conversation agent. This enables analytics for user prompts, agent responses, and conversation restarts.

Changes

• Added pendo.trackAgent('prompt', ...) call in handleSendMessage to track when users submit messages, including the message content and a generated messageId
• Added pendo.trackAgent('agent_response', ...) call after successful API response to track assistant replies with the response content
• Added pendo.trackAgent('user_reaction', ...) call with content 'retry' in handleRestart to track when users restart conversations
• Added TypeScript declare statement for the global pendo object with trackAgent method signature
• All trackAgent calls are guarded with typeof pendo !== 'undefined' checks to prevent runtime errors if Pendo is not loaded

Notes

• The agentId is an empty string as specified in the agent configuration
• No thumbs up/down feedback UI exists in this component, so user_reaction tracking is limited to the restart/retry action
• messageIds are generated using crypto.randomUUID() for each event

---

Generated by Novus.