Intent-driven API debugging UI powered by Generative AI
IntentTrace is a developer-focused debugging tool that replaces dashboards, logs, and guesswork with an intent-first interface. Instead of navigating tools manually, developers describe their API issue in plain English β and the UI dynamically assembles the right debugging components to investigate the problem.
This is not a chatbot.
This is Generative UI for debugging.
π Try IntentTrace live:
https://intent-trace.vercel.app
Describe an API issue in plain English and see the UI assemble the right debugging tools instantly.
Traditional tools:
- Show everything
- Require manual navigation
- Force developers to know where to look
IntentTrace:
- Starts with intent
- Shows only relevant debugging tools
- Hides noise
- Adapts the UI per problem
βDescribe the problem β Get the tools you needβ
-
Developer types a debugging problem
Example:- βJWT expiredβ
- βWhich APIs are slow?β
- βCORS error in browser but works in Postmanβ
-
AI analyzes the intent
-
Based on the intent, the system:
- Selects relevant debugging components
- Renders them directly in the UI
-
No chat UI, no explanations unless needed β just tools.
- AI decides which debugging cards to show
- No hardcoded conditions
- No manual toggles
AuthFlowCard
- Visualizes login β token β protected API flow
- Clearly highlights failures (e.g. expired tokens, auth issues)
- Renders only when auth-related intent is detected
JWTDecoderCard
- Decodes JWT header & payload
- Detects token expiration
- Explains why authentication fails
- Never renders unless JWT-related intent exists
RequestResponseCard
- Displays HTTP method, endpoint, status, and error message
- Useful for 401s, 500s, and failed API calls
LatencyChartCard
- Highlights slow APIs
- Shows response times
- Renders only for performance-related issues
EndpointListCard
- Lists affected endpoints
- Shows which APIs are slow and by how much
ErrorInsightCard
- Explains possible reasons behind failures
- Helps developers decide next steps
- Only shown when error context exists
UI renders:
- Authentication Flow
- JWT Token Inspector
UI renders:
- Performance Overview
- Affected Endpoints
UI renders:
- Request / Response Debug
- Error Insights
No unnecessary cards.
No irrelevant tools.
- React + Vite
- Tambo AI (Generative UI & intent-based rendering)
- Zod (component prop schemas)
- Modern CSS (dark developer-first UI)
src/
βββ app/
β βββ Layout.jsx # Application layout wrapper
β
βββ components/
β βββ core/ # Core application flow
β β βββ DynamicCanvas.jsx # Intent-driven UI renderer (Tambo output)
β β βββ EmptyState.jsx # Initial state with example prompts
β β βββ PromptInput.jsx # Natural language input interface
β β
β βββ debug/ # Intent-driven debugging UI components
β β βββ AuthFlowCard.jsx # Authentication flow & token failure analysis
β β βββ CorsInsightCard.jsx # CORS error detection & browser/Postman mismatch
β β βββ EndpointListCard.jsx # Affected API endpoints with response times
β β βββ ErrorInsightCard.jsx # Root-cause explanations and debugging insights
β β βββ JWTDecoderCard.jsx # JWT token inspection & expiry detection
β β βββ LatencyChartCard.jsx # API latency & performance visualization
β β βββ RequestResponseCard.jsx # HTTP request / response debugging
β β
β βββ tambo/ # Tambo UI & SDK components
β β βββ base/ # Internal Tambo rendering primitives
β β βββ control-bar.tsx # Command / interaction UI
β β βββ message-input.tsx # Prompt input for AI threads
β β βββ message.tsx # AI message rendering
β β βββ thread-content.tsx # Conversation thread container
β β βββ ... # Supporting Tambo components
β β
β βββ ui/ # Reusable UI primitives
β βββ Badge.jsx # Status & label badges
β βββ Card.jsx # Base card container
β βββ CodeBlock.jsx # Syntax-highlighted code blocks
β βββ Divider.jsx # Visual separators
β
βββ data/ # Mock debugging data
β βββ mockLogs.js # Simulated API logs
β βββ mockMetrics.js # Performance & latency metrics
β βββ mockRequests.js # Sample HTTP requests
β
βββ intents/ # Intent classification & mapping
β βββ intentMap.js # Intent β component selection logic
β βββ samplePrompts.js # Example user debugging prompts
β
βββ tambo/ # Generative UI integration layer
β βββ registry.ts # Central registry for all Tambo components
β βββ thread-hooks.ts # Hooks for managing AI conversation threads
β βββ utils.ts # Shared helpers for Tambo integration
β
βββ styles/ # Styling & theming
β βββ globals.css # Global styles
β βββ theme.css # Theme variables (dark developer UI)
β
βββ utils/ # General helper utilities
β βββ formatters.js # Data formatting helpers
β βββ helpers.js # Miscellaneous utilities
β
βββ App.jsx # Root application component
βββ main.jsx # Application entry point
All debugging components are registered centrally in tambo.ts with:
- Name
- Description
- Intent guidance
- Zod
propsSchema
This allows the AI to:
- Understand when to render a component
- Avoid rendering irrelevant tools
- Keep the UI clean and focused
IntentTrace demonstrates a new UX pattern:
AI doesnβt answer questions β it assembles interfaces.
This approach:
- Reduces cognitive load
- Speeds up debugging
- Makes complex systems easier to reason about
β
Core functionality complete
β
Stable generative UI behavior
β
Demo-ready for hackathons / showcases
MIT
Built for developers who debug under pressure.