Release Notes

🎉 New Features

1. Feedback Module

User Feedback Collection System

A comprehensive feedback collection system that captures user sentiment and suggestions to improve the application.

Key Features:

Multi-Dimensional Feedback Collection
- Emotion-based rating system with 5 levels (Terrible, Bad, Okay, Good, Amazing)
- Net Promoter Score (NPS) rating from 1-10
- Free-form text comments/suggestions (up to 1000 characters)
- Optional email capture for follow-up
Smart Display Logic
- Automatically appears after 5 minutes of user activity
- 24-hour cooldown between displays to avoid survey fatigue
- Dismissing the modal counts as a submission to prevent repeated prompts
- Persistent tracking using localStorage
User Experience
- Interactive emoji-based emotion selector with grayscale/color transition effects
- Color-coded NPS buttons (red for detractors 1-6, yellow for passives 7-8, green for promoters 9-10)
- Responsive design optimized for mobile and desktop
- Auto-populated email for authenticated users
- Real-time character counter for feedback messages
Data Management
- Integrates with Google Sheets for centralized feedback storage
- Captures: timestamp, user ID, email, emotion rating, NPS score, and message
- Type-safe validation using Zod schema
- Toast notifications for submission success/failure

Technical Implementation:

Components: feedback-modal.tsx
API: /api/feedback (POST endpoint)
Services: feedback.services.ts (Google Sheets integration)
Types: feedback.types.ts (EmotionRating enum, FeedbackSchema, FeedbackInput, FeedbackResponse)
Tanstack Query: Mutation hook for optimistic updates and error handling
Storage: LocalStorage for submission tracking

2. Financial Analyst AI Agent

Intelligent Conversational AI for Financial Data Analysis

A read-only AI agent powered by GPT-4o-mini that helps users understand their financial data through natural language conversations.

Key Capabilities:

Query & Analysis:

Analyze spending patterns and income sources
Provide financial summaries (income, expenses, balance)
Query transactions by date, category, or wallet
Compare spending across different time periods
Show recent activity and transaction history
List and describe available wallets and categories
Validate entity existence

Natural Language Date Processing:

Understands relative dates ("last month", "this year", "yesterday")
Parses absolute dates ("January 22, 2026", "2026-01-22", "01/22/2026")
Automatically gets current date/time for context-aware responses

Smart Features:

Queries across ALL wallets by default unless user specifies otherwise
Multi-step reasoning with up to 15 tool call iterations
Real-time transaction amount formatting (Philippine Peso ₱ with 2 decimal places)
Conversational, friendly responses with financial insights
Pattern detection and recommendations

Tool Arsenal (21 Read-Only Tools):

Date & Time Tools (3):

getCurrentDateTime - Get current date and time
parseRelativeDate - Parse relative expressions (e.g., "last month")
parseAbsoluteDate - Parse specific dates (e.g., "January 22, 2026")

Transaction Tools (5):

getTransactions - Paginated transactions with filters
getTransactionById - Specific transaction details
getRecentTransactions - Recent activity
getTransactionsByDateRange - Date range queries
getTransactionSummary - Financial summary (income, expense, balance)

Wallet Tools (4):

getWallets - All user wallets with balances
getWalletsPaginated - Paginated wallets with search
getWalletById - Specific wallet details
getWalletBalance - Wallet balance calculation

Category Tools (4):

getCategories - All available categories
getCategoriesByScope - Filter by GLOBAL/USER scope
getCategoryById - Specific category details
getCategoryCountByType - Category statistics

User & Validation Tools (5):

getUserStats - User statistics overview
getUserProfile - User profile details
walletExists - Validate wallet existence
categoryExists - Validate category existence
transactionExists - Validate transaction existence
getAuthenticatedUserId - Extract current user context

User Interface:

Floating Chatbot Widget
- Fixed position in bottom-right corner
- Expandable/collapsible with minimize/maximize controls
- Responsive design (full-screen on mobile, fixed 600px height on desktop)
- Real-time streaming responses
- Loading indicators and error states
- Markdown support for formatted responses
- Experimental feature warning badge

Security & Performance:

Rate Limiting:

100 requests per user per 60-minute window
In-memory tracking with automatic cleanup
Returns remaining quota and reset timestamp
HTTP 429 responses with proper headers when exceeded

Authentication:

Requires authenticated user session
HTTP 401 responses for unauthorized access
User context automatically passed to agent

Error Handling:

Safe tool execution with try-catch wrappers
Graceful degradation for missing data
Clear error messages to users
Comprehensive logging for debugging

Technical Implementation:

Agent: financial-analyst-agent.ts (ToolLoopAgent with GPT-4o-mini)
API: /api/chat (POST endpoint with streaming)
Components: floating-chatbot.tsx
Tools: Organized by category (date, transaction, wallet, category, user)
Utilities: Rate limiter, response formatters
Types: Type-safe message inference with FinancialAnalystAgentMessage
Framework: Vercel AI SDK with streaming support
State Management: React hooks (useChat)

🔧 Technical Notes

Dependencies:

AI/ML: @ai-sdk/openai, @ai-sdk/react, ai
Validation: zod
HTTP Client: Native fetch
State Management: @tanstack/react-query
UI Components: Shadcn/ui components (Dialog, Button, Input, Textarea, Card, etc.)
Google Integration: Google Sheets API for feedback storage

Environment Requirements:

OpenAI API key (for GPT-4o-mini)
Google Sheets API credentials (for feedback storage)
Authenticated user session (Supabase Auth)

Performance Considerations:

Rate limiting prevents API abuse
In-memory caching for rate limit data
Streaming responses for better UX
Automatic cleanup of expired rate limit entries
Tool result logging for monitoring

📝 Usage Examples

Feedback Module:

Users will automatically see the feedback modal after:

Being active in the app for 5 minutes
At least 24 hours have passed since last submission/dismissal

AI Agent Examples:

"What did I spend last month?"
"Show my income for this year"
"How much is in my Cash wallet?"
"List my top 5 expenses in January"
"Compare my spending this month vs last month"
"What categories am I spending the most on?"

⚠️ Known Limitations

Feedback Module:

Requires client-side JavaScript
LocalStorage-based tracking (cleared if user clears browser data)
Google Sheets dependency for storage

AI Agent:

Read-only: Cannot create, update, or delete data
Rate limited to 100 requests/hour per user
Experimental feature - may provide inaccurate information
Requires active internet connection for API calls
Tool execution limited to 15 steps per conversation
No conversation history persistence (stateless)

🚀 Future Enhancements

AI Agent:

Conversation history persistence
Export conversation to PDF/Email
Voice input/output support
Proactive financial insights and alerts
Budget recommendations
Spending anomaly detection
Write capabilities (create transactions, set budgets)

What's Changed

[release] v2.0 by @npeteros in #8
Release: Deploy version 2.0 to Production by @npeteros in #10
feat: enable user redirection to login page when no session is found by @npeteros in #11
fix: ensure proper redirection to login page when no user session is … by @npeteros in #12
refactor: remove console logs and update redirect path in authentication components by @npeteros in #13
feat: implement session management with API integration for user authentication by @npeteros in #14
refactor: consolidate user data retrieval by updating user-related hooks and componenets by @npeteros in #15
refactor: remove unused user data retrieval and implement logout functionality in dashboard layout by @npeteros in #16
refactor: added metadata for SEO by @npeteros in #17

New Contributors

@npeteros made their first contribution in #3

Full Changelog: https://github.com/npeteros/audit/commits/v2.0

Releases: npeteros/audit

Release list

v2.1.0

Release Notes

🎉 New Features

1. Feedback Module

Key Features:

Technical Implementation:

2. Financial Analyst AI Agent

Key Capabilities:

Tool Arsenal (21 Read-Only Tools):

User Interface:

Security & Performance:

Technical Implementation:

🔧 Technical Notes

Dependencies:

Environment Requirements:

Performance Considerations:

📝 Usage Examples

Feedback Module:

AI Agent Examples:

⚠️ Known Limitations

Feedback Module:

AI Agent:

🚀 Future Enhancements

AI Agent:

Uh oh!

v2.0.1

What's Changed

Contributors

Uh oh!

v2.0

What's Changed

New Contributors

Contributors

Uh oh!