Reading the repo — stats and the shipped feed land in about ten seconds…
2 0 0 openLast shipped 67d ago
A deterministic 0–100 hygiene score — README, license, CI, tests, docs, and freshness.
Who ships this repo — author concentration and the bus factor across the last 300 mainline commits.
How welcoming this repo is to contributors — issue throughput, close time, responsiveness, and good-first-issue count.
Grounded in tokenomics's README, structure, and recent commits — answers won't invent code they haven't seen.
A Monday email with what shipped, in plain English — no account needed.
Major codebase restructure and tooling
Reorganized the entire project into a modular architecture with new core services (session indexing, parser registry, security, caching), added test suite with unit tests, implemented code formatting with Prettier, and removed legacy backup files. New routes for models and sessions were added alongside enhanced security and storage layers.
Add session list views
Introduced new views for displaying sessions with enhanced UI styling and layout options. Users can now browse and view session information in list format with improved visual presentation.
Fix session filtering and deduplication
Fixed multiple context bugs: empty placeholder sessions are now hidden from the sidebar, duplicate multi-root workspace sessions are deduplicated, removed sessions properly trigger cleanup, and the app now prefers newer chatSessions format over older debug-logs format. Recent sessions are visually emphasized while older ones fade into the background.
Support dynamic context window sizes
The app now handles different context window sizes per model instead of assuming a fixed 200K limit. Claude models default to 200K, while Codex models use 258K, with the window size displayed throughout the UI and used in all token calculations.
Fix token usage deduplication
Improved token usage tracking for Codex CLI/VS Code format by deduplicating repeated events based on total token counts, and fixed context size calculation in Copilot parser to use actual input tokens instead of cumulative values. Also enhanced session detection to properly locate Codex files in subdirectories and extract project names from metadata.
Add API key management UI
Users can now set and update their Anthropic API key directly in settings, with the key securely stored in config.json and only a masked preview shown in the interface. The setting is used for AI audit and analysis features.
Added time series page and navigation
New time series visualization page with updated site navigation and styling. The app now includes dedicated routing for time series data analysis alongside refreshed UI components.
Reorganize codebase into modular architecture
The backend has been refactored from a single monolithic server.js file into a modular structure with separate concerns: routing (router.js), AI integrations (src/ai.js, parsers for Claude/Codex/Copilot), features (analytics, history, notifications), and individual route handlers. The frontend also gained an alert drawer UI for viewing all active alerts across sessions in one place, with improved visual indicators (warning icons) on session items.
Fix Windows Copilot detection
Added cross-platform support for detecting GitHub Copilot chat sessions, with proper Windows (APPDATA), macOS (Library), and Linux (.config) directory paths. Also improved chat session parsing to handle multiple JSON object formats and extract token usage data more reliably.
Initial app release
Launched the first version of Repo Wrapped with a complete frontend interface (HTML, CSS, JavaScript) and backend server setup. Users can now access the application through a web browser.
A floor, not a guess: counts only commits whose author, co-author trailer, or message explicitly credits an AI tool (Claude, Copilot, Cursor, aider, Codex…). Based on 11 mainline commits. Unattributed AI code isn't counted here — the full audit estimates that separately.
