Usage Guide

Dashboard

The main dashboard uses a customizable widget grid layout. Enter Edit Mode to add, remove, resize, and reorder widgets.

Available Widgets

Widget Sizes

• S (Small) — 1 column
• M (Medium) — 2 columns
• L (Large) — 4 columns, full width

Dashboard Controls

• Time Range — Select from presets (15m, 30m, 1h, 3h, 6h, 12h, 24h, 3d, 7d) or set a custom date-time range. All time-based widgets update together.
• Chart Fill — Choose between Line, Gradient, or Solid fill styles for charts.
• Refresh — Manually refresh all widget data.
• Edit Mode — Toggle to add/remove/resize widgets. Each widget shows S/M/L size buttons and a remove button.

Ask AI

Natural language Q&A over your VectorWave data using RAG.

How to Use

1. Navigate to Ask AI in the sidebar
2. Select a function scope — choose "All Functions" or a specific function to narrow the context
3. Type your question in the text area (e.g., "Why is my embed function failing?")
4. Click Ask

The AI retrieves relevant function data, execution logs, and error context from Weaviate, then generates an answer.

Plan Limits

Free plan users see a usage bar showing today's remaining calls. When the daily limit is reached, the Ask button is disabled with a message: "Daily limit reached. Upgrade to Pro for unlimited access."

Saving and Bookmarking

Responses from Ask AI are automatically saved to your AI History. You can bookmark important responses by clicking the bookmark icon on any result.

AI History (Saved Responses)

Browse and manage all AI responses from Ask AI and Healer.

Filtering

• Source filter — Show all, Ask AI only, or Healer only
• Bookmarked — Toggle to show only bookmarked responses
• Search — Full-text search across questions and answers

Response Cards

Each card shows the source icon, question preview, source type badge, function name (if scoped), timestamp, and bookmark status. Click to expand and view the full answer.

Actions

• Bookmark / Unbookmark — Toggle bookmark status on any response
• Delete — Remove a response permanently (with confirmation)

Free Plan Limitation

On the Free plan, responses older than 24 hours are locked (shown with reduced opacity and a lock icon). Upgrade to Pro for unlimited access to your full AI history.

Trace Waterfall

The trace waterfall visualizes distributed call chains as interactive diagrams.

Reading the Waterfall

pipeline()                    ████████████████████████░░░░░ 450ms
├── embed()                   ██████████░░░░░░░░░░░░░░░░░░  180ms
├── search()                  ░░░░░░░░░░████████░░░░░░░░░░  150ms
└── format_result()           ░░░░░░░░░░░░░░░░░░██░░░░░░░░   20ms

• Horizontal bars represent span duration
• Nesting shows parent-child relationships
• Colors indicate status:
  • Green = success
  • Red = error
  • Blue = cache hit
  • Gray = in progress

Span Details

Click any span to see:

• Function name and module path
• Input parameters and return value
• Error details and stack trace (if failed)
• Duration breakdown and cache hit status

Filtering Traces

• Filter by function name, status, or duration range
• Sort by timestamp, duration, or error count
• Search traces by trace ID

AI Healer

The AI Healer diagnoses errors and suggests code fixes. Requires an OpenAI API key.

Single Diagnosis

1. Select a function from the dropdown
2. Adjust the lookback slider to set the time window
3. Click Diagnose & Heal
4. View the diagnosis: root cause analysis and suggested fix (syntax-highlighted Python code)
5. Copy the fix or Save the diagnosis to AI History (Pro only)

Batch Diagnosis

1. Switch to Batch Mode
2. Select multiple functions using checkboxes
3. Click Batch Diagnose (N)
4. Review all diagnoses in a summary view

Time Range

Choose a lookback window: 1h, 3h, 6h, 12h, 24h, 3d, 7d, or All.

Replay Testing

Visual regression testing for AI functions.

Running a Replay

1. Select a function from the sidebar
2. Choose test entries:
   • Golden Only — Verified Golden Dataset entries
   • All — All stored executions
3. Select comparison mode:
   • Exact Match — Character-by-character comparison
   • Semantic — Embedding similarity with configurable threshold
4. Click Run Replay

Results View

• Summary — Pass/fail counts, success rate
• Detail View — Side-by-side comparison of expected vs actual output
• Similarity Score — Cosine similarity for semantic comparisons
• Diff View — Character-level diff for exact match failures

Golden Dataset Management

• Browse all Golden Dataset entries per function
• Promote new executions to Golden status
• Remove outdated entries
• View AI-recommended candidates for promotion

Semantic Search

Full-text and vector search across all your AI function data.

Search Interface

• Natural language search bar at the top
• Real-time results as you type
• Filter by function name, status, date range, or error code

Search Results

Results include:

• Functions — Matching function definitions
• Executions — Matching execution logs with input/output previews
• Errors — Matching error logs with stack traces
• Traces — Matching trace chains with waterfall preview

Settings

Appearance

Three theme options: Dark (neon dark mode), Light (clean light mode), System (follows OS preference).

AI Settings

Configure your global OpenAI API key for all AI features.

Connections

Manage Weaviate connections. Set per-connection OpenAI API keys for isolated AI access.

Account

View your profile, plan status, and security settings. Upgrade or downgrade your plan.

Multi-Language UI

VectorSurfer supports three languages:

Switch languages from the login page, signup page, projects page, or the account settings. All UI elements, labels, and messages are translated.

Next Steps

• Getting Started — Connect to VectorSurfer
• VectorWave Overview — The core SDK
• VectorSurferSTL — Lightweight alternative