RequestTracker

RequestTracker is an Avalonia UI desktop application designed to visualize and analyze session logs and request data from AI coding assistants like GitHub Copilot and Cursor. It provides a unified view of interactions, enabling developers to review prompt history, context usage, and automated actions taken by these tools.

Key Features

• Unified Dashboard: Aggregates logs from different AI providers (Copilot, Cursor) into a single, searchable interface.
• Tree Navigation: filesystem watcher monitors a target directory for log files, automatically indexing them into a navigation tree.
• Detailed Request Analysis:
  • Interpretation & Metadata: Displays extracted metadata, user intent interpretation, and key decisions.
  • Context Inspection: View the exact context (code snippets, files) sent to the LLM.
  • Action Tracking: visualizes automated actions (file creation, edits, command execution) in a structured grid.
  • JSON Inspector: Built-in JSON viewer to inspect the raw underlying data for deep debugging.
• Markdown Rendering: Integrated markdig-based markdown viewer for rendering log content and associated documentation.
• Responsive UI: Modern, light-theme interface with collapsible sections and persistent window state.

Data Ingestion

The application monitors a specified directory (e.g., docs/requests) for JSON and Markdown log files. It supports custom schema formats used to track AI interactions:

• Copilot Logs: JSON-based session logs containing request/response cycles, token usage metrics, and workspace context.
• Cursor Logs: JSON/Markdown hybrid logs capturing "Tab" requests, diffs, and chat history.

Usage

1. Configure Directory: Point the application to the root folder containing your request logs.
2. Browse Sessions: Use the tree view on the left to navigate through sessions organized by date or folder structure.
3. Inspect Requests: Click on a specific request to view its details in the main panel.
   • Expand the Actions grid to see what file changes were performed.
   • Check Interpretation to understand how the AI understood the task.
   • Use the Original JSON expander to see the raw data fields.

Configuration

Edit src/RequestTracker/appsettings.config (copied next to the executable) to set Mcp.BaseUrl, Paths.SessionsRootPath, Paths.HtmlCacheDirectory, and Paths.CssFallbackPath. Session logs are loaded from Mcp.BaseUrl (/mcp/sessionlog), while Paths.* is still used for documents/source context and markdown CSS fallback. Paths.HtmlCacheDirectory supports environment variables such as %TEMP%.

Tech Stack

• Framework: Avalonia UI (Cross-platform .NET XAML framework)
• Language: C# / .NET 8
• Parsing: System.Text.Json with custom robust parsing for flexible schemas.
• Markdown: Markdig.Avalonia for rendering formatted text.

Build and run

# From repo root
cd src/RequestTracker
dotnet run

WSL with WSLg

On WSL with WSLg enabled (Windows 11), the app window should appear on the Windows desktop. If it doesn’t:

1. Check WSLg: Ensure you’re on Windows 11 with WSLg (no separate X server needed).
2. Run from project: cd src/RequestTracker && dotnet run -c Debug
3. Or use the script: From repo root, chmod +x run-wslg.sh && ./run-wslg.sh
4. Taskbar: The window may show in the Windows taskbar; click it to bring it to front.