GitDash

A personal GitHub activity dashboard with gamified tier progression and AI-powered commit summaries.

Created by shuxueshuxue.

Features

Tier-Based Progression System

• 12 tiers from "slacker" to "transcendent"
• Dynamic scoring: weight = 5 × (3-day commits) + (30-day commits)
• Color-coded progress bar showing advancement to next tier
• Real-time activity score with glowing tier labels

Activity Dashboard

• Weight-based ranking of repositories (not just by recency)
• Color-coded commit counts by activity level:
  • Red (50+) - Critical activity
  • Orange (25-49) - High activity
  • Green (10-24) - Medium activity
  • Blue (1-9) - Low activity
  • Gray (0) - Inactive
• Language icons instead of full names
• AI-generated working state summaries (concise keywords)

Smart Caching & Refresh

• Backend auto-refreshes from GitHub every 60 seconds
• Frontend polls backend every 60 seconds (lightweight)
• Intelligent cache: only fetches commits for updated repos
• 90% API call reduction vs naive approach
• Manual refresh button for instant updates

Dark Retrospective UI

• Monospace font, terminal-inspired design
• Minimal color palette for focus
• Rainbow gradient score display
• Smooth animations and transitions

Installation

Prerequisites

• Python 3.12+
• uv package manager
• GitHub Personal Access Token

Setup

1. Clone the repository

   git clone https://github.com/shuxueshuxue/gitDash.git
   cd gitDash

2. Install dependencies

   uv sync

3. Create GitHub Token

   • Go to https://github.com/settings/tokens
   • Generate new token (fine-grained recommended)
   • Required permissions: Read-only access to "Metadata" and "Commit statuses"

4. Configure API keys

   Copy the example config and fill in your credentials:

   cp models.json.example models.json

   Example format:

   {
     "models": {
       "claude-haiku-4.5": {
         "endpoint": "YOUR_API_ENDPOINT",
         "api_key": "YOUR_API_KEY",
         "model": "anthropic/claude-haiku-4.5"
       }
     },
     "default_retry": {
       "max_retries": 3,
       "timeout": 600
     }
   }

5. Set environment variable

   export GITHUB_TOKEN="your_github_token_here"

Usage

CLI Mode

uv run python main.py

Web Dashboard

uv run python server.py

Then open http://localhost:8000 in your browser.

Tier System

The tier system ranks your coding activity across your top 15 repositories:

Score Range	Tier	Color
0-49	slacker	Dark Gray
50-99	dabbler	Gray
100-199	apprentice	Blue
200-299	craftsman	Light Green
300-399	specialist	Green
400-499	master	Lime
500-749	prodigy	Yellow
750-999	legend	Orange
1000-1499	machine	Light Red
1500-1999	demigod	Red
2000-2999	deity	Deep Red
3000+	transcendent	Intense Red

Score Calculation:

• Each repo gets: 5 × (commits in last 3 days) + (commits in last 30 days)
• Total score = sum of all repo scores
• Recent activity (3 days) weighted 5× more than older activity

Architecture

Components

• GitHubClient - GitHub REST API wrapper with rate limiting
• CommitAgent - Smart caching + AI summary generation (uses PolyAgent)
• Board - Pure computation: metrics & ranking
• server.py - FastAPI web server with background refresh task

Data Flow

GitHub API → CommitAgent (cache + AI) → Board (compute) → Frontend
     ↑                                                         ↓
     └─────────── Background Task (60s) ──────────────────────┘

Features

• Transparent caching: Board doesn't know about cache existence
• Time-travel debugging: Set as_of parameter to test different dates
• Incremental fetching: Only updates repos with new commits
• Structured AI output: Pydantic schemas for consistent summaries

Configuration

Adjusting Refresh Interval

Edit server.py:

await asyncio.sleep(60)  # Change to desired seconds

Changing Repo Limit

Edit server.py:

await refresh_dashboard(limit=15)  # Change limit

Customizing Tiers

Edit the TIERS array in server.py to adjust ranges and names.

Tech Stack

• Backend: FastAPI, AsyncIO
• GitHub API: httpx, GitHub REST API v3
• AI: PolyAgent (multi-provider LLM client)
• Data: JSON-based caching (no database needed)
• Frontend: Pure HTML/CSS/JS (no framework)

License

MIT

Credits

Built with PolyAgent for AI-powered commit summaries.