Turn GitHub Copilot usage into AI-powered executive insight. The new AI Analyst converts your enterprise Copilot data into grounded recommendations, while the dashboard gives engineering leaders full visibility into adoption, licensing costs, AI credit spend, and AI model activity across every team and organization.
Copilot Insights now includes an optional AI Analyst powered by the GitHub Copilot SDK. It turns your synced Copilot data into structured Markdown briefings for engineering leaders, covering cost and license optimization, adoption coaching, executive summaries, delivery impact, ROI and spend forecasts, and team scorecards.
The AI Analyst is designed for transparency and governance:
- 🧠 Grounded analysis — every narrative is generated from dashboard metrics, enterprise context, org topology, feature mix, and persisted access/seat signals
- 📊 Business-ready output — structured Markdown with key metrics, risks, recommended actions, caveats, and analysis confidence
- 🏢 Enterprise personalization — admins can provide additional instructions and assumptions, such as hourly cost, priority teams, and reporting preferences
- 🔒 Operational control — AI is off by default, uses a dedicated Copilot token, caches generated insights, and includes a cache management panel
- 🔍 Transparency report — see docs/ai-analyst.md for what data is sent, prompt structure, cache behavior, and security model
GitHub Copilot is transforming how teams write code — but without visibility into how it's being used, it's hard to measure the return on your investment. Copilot Insights bridges that gap by providing:
- 📈 Adoption tracking — see which teams and users are actively using Copilot, and where adoption lags
- 💰 License optimization — identify unused seats and potential savings across your enterprise
- 🤖 Model intelligence — understand which AI models drive the most value and how AI credit budgets are consumed
- 🧠 AI-powered analyst briefings — convert raw metrics into executive, cost, adoption, delivery, ROI, and team recommendations
- ⚡ Productivity metrics — measure code completions, agent usage, PR impact, and CLI adoption in one place
- 🔒 Enterprise governance — role-based access, audit logging, and team-level filtering for compliance
| Page | Route | Description |
|---|---|---|
| AI Analyst | /ai-analyst |
AI-generated executive, cost, adoption, delivery, ROI, and team analysis grounded in synced dashboard data |
| Copilot Usage | /metrics |
Daily/weekly active users, code completions, chat mode breakdown, model & language analytics |
| Code Generation | /code-generation |
LOC added/deleted by user vs agent, breakdowns by feature, model, and language |
| PR & Autofix | /pull-requests |
AI-assisted PR creation, Copilot code review suggestions, autofix analytics, and merge metrics |
| Agent Impact | /agents |
Agent adoption rate, IDE Agent vs GitHub Coding Agent breakdown, top agent users |
| AI Adoption | /ai-adoption |
Cohort-based adoption analysis (code-first, agent-first, multi-agent), progression trends, and per-cohort productivity |
| CLI Impact | /cli |
CLI adoption, session/request volumes, token consumption, version distribution |
| Copilot Licensing | /seats |
Seat assignments, license utilization, plan distribution, savings opportunities (live from GitHub API) |
| AI Credits | /ai-credits |
AI credit consumption, included pool utilization, model/org/team/user breakdowns, and monthly trends |
| Premium Requests | /premium-requests |
Historical premium-request usage report retained for pre-usage-based-billing periods |
| Models | /models |
AI model catalog with usage stats, premium vs included tiers, and feature breakdown |
| Users | /users |
Individual user activity explorer with license status, engagement patterns, and feature adoption |
| Enterprise Teams | /enterprise-teams |
Team management with member sync from GitHub Enterprise Teams API |
| Metrics Reference | /reference |
200+ metric definitions, calculation formulas, and data sources |
- 🌍 Internationalization — 7 languages (English, Arabic RTL, Spanish, French, German, Hindi, Italian) via
useTranslation()hook - 🌓 Dark/Light/System theme — three-mode theme with
dark:Tailwind variants and theme-aware Chart.js options - 📄 PDF export — one-click PDF generation for all dashboard pages
- 🔎 Multi-select filters — filter charts by organization, enterprise team, user, model, or language
- 🧠 AI Analyst — optional Copilot SDK-powered analysis with grounded prompts, Markdown output, caching, admin assumptions, and enterprise context
- 🚩 Configuration banner — shown when GitHub token or enterprise slug is missing
- ℹ️ About report context — each report includes an “About this report” banner with metric scope/context
- 📝 Audit logging — tracks admin actions for compliance
- 🔐 Dashboard auth gate — optional password protection for all dashboard pages
Welcome page with feature overview cards and quick navigation to all dashboards.
AI-generated executive, ROI, and team scorecard analysis with grounded Markdown summaries and collapsible cards.
Daily and weekly active users, code completions, chat requests, and model usage trends.
Lines of code added and deleted across modes, models, and languages.
AI-assisted pull request creation, Copilot code review suggestions, and autofix analytics.
Agent adoption rate, IDE Agent vs GitHub Coding Agent breakdown, and top agent users.
User adoption cohorts (code-first, agent-first, multi-agent), progression over time, and per-cohort productivity.
GitHub Copilot CLI adoption, session and request volumes, and token consumption.
License utilization, seat costs, and savings opportunities — live from GitHub API.
AI credit usage, included credit pool utilization, and billing breakdowns by model, org, team, and user.
Premium request consumption report for historical periods before AI credit billing took effect.
Individual user activity explorer with license status, engagement metrics, and feature adoption.
Enterprise team management with member sync from GitHub Enterprise Teams API.
200+ metric definitions with calculation formulas, data sources, and usage guidance.
Manage your GitHub connection and enterprise slug.
Configure the AI Analyst token, model, admin assumptions, enterprise context status, and cached insight management.
Schedule automatic syncs, trigger manual pulls, or upload NDJSON exports.
- Frontend: Next.js 16.2 App Router, React 19.2, Tailwind CSS 4.2, Chart.js 4.5
- Backend: Next.js API routes, Drizzle ORM 0.45, Zod 4.3
- Database: PostgreSQL 18 (star schema — 10 dimensions + 9 fact tables)
- ETL: Custom ingest pipeline with GitHub Copilot Usage Metrics API (v2026-03-10)
- Billing data: Seats fetched live; AI Credits fetched from GitHub billing API and snapshotted for trend continuity
- Infrastructure: Azure Container Apps, Azure Database for PostgreSQL, Azure Container Registry, Key Vault
See docs/architecture.md for detailed architecture documentation.
The optional AI Analyst turns your Copilot data into written executive, cost, adoption, and delivery analysis using the GitHub Copilot SDK. See docs/ai-analyst.md for a full transparency report on the data it gathers, the prompts it uses, and how analysis is generated, cached, and secured.
- Node.js 24+ and pnpm (
corepack enable) - PostgreSQL 18+ (local or cloud)
- GitHub Enterprise Cloud with Copilot enabled
- GitHub Personal Access Token with
manage_billing:copilot,read:enterprise,read:orgscopes
# 1. Clone the repository
git clone https://github.com/your-org/ghcp-dashboard.git
cd ghcp-dashboard
# 2. Install dependencies
cd app
pnpm install
# 3. Configure environment
cp .env.example .env
# Edit .env with your database URL and admin password
# 4. Run database migrations
pnpm exec drizzle-kit migrate
# 5. Start the development server
pnpm run devOpen http://localhost:3000 and navigate to Settings to configure your GitHub token and sync schedule.
| Variable | Required | Description |
|---|---|---|
DATABASE_URL |
Yes | PostgreSQL connection string (e.g. postgresql://user:pass@host:5432/db) |
ADMIN_PASSWORD |
No | Password for Settings page access. If unset, settings are open to all. Only leave this unset for local development on a trusted machine |
DASHBOARD_PASSWORD |
No | Password gate for all dashboard pages. If unset, dashboards are open to all. Only leave this unset for local development or intentionally public dashboards |
NEXT_PUBLIC_BUILD_ID |
No | Git commit SHA shown in sidebar footer (auto-set in Docker) |
NEXT_PUBLIC_BUILD_TIME |
No | Build timestamp shown in sidebar footer (auto-set in Docker) |
Warning
For any non-local deployment, set ADMIN_PASSWORD to protect the Settings page. The Settings UI can configure the GitHub token, Enterprise slug, and sync interval. If dashboard pages should not be publicly accessible, also set DASHBOARD_PASSWORD. Leaving these unset is only recommended for local development on a trusted machine.
The GitHub token, Enterprise slug, and sync interval are configured via the Settings UI and stored in the database.
Azure is the recommended — and simplest — way to get the dashboard up and running. A single azd up provisions every resource, wires up secrets and managed identity, and deploys the app. If you'd rather run it yourself, a self-hosted option is fully supported too.
The fastest path to a production-ready deployment. This project ships Infrastructure as Code (Bicep) and an Azure Developer CLI configuration, so one command stands everything up for you (Container Apps, PostgreSQL, Container Registry, Key Vault, and Application Insights).
Requirements
- An Azure subscription
- Azure Developer CLI (
azd) - Docker (used to build the app image)
Deploy
# Provision infrastructure + deploy the app (first run)
azd up
# Subsequent deploys (app only)
azd deployWhen the deploy finishes, azd prints the app URL. Open it, go to Settings, and add your GitHub token and sync schedule — that's it.
A typical deployment in the North Europe region costs approximately $45 / month (USD), ranging from roughly $35 to $65 / month depending on dashboard traffic and log volume. Figures are pay-as-you-go list-price estimates; reserved capacity or savings plans reduce them further. For a precise quote, use the Azure Pricing Calculator.
Run the dashboard anywhere you can host a container — your own VM, Kubernetes, another cloud, or on-premises.
Requirements
- PostgreSQL 18+ (managed or self-run), reachable from the app
- Docker — the app ships a production
Dockerfilewith Next.js standalone output (or use Node.js 24+ and pnpm for local development) - TLS / reverse proxy (e.g. nginx, Caddy) in front of the app for any shared deployment
- GitHub Personal Access Token with
manage_billing:copilot,read:enterprise,read:org(added later via the Settings UI) - Environment variables (see Environment Variables):
DATABASE_URL— requiredADMIN_PASSWORD— required for any shared/non-local deploymentDASHBOARD_PASSWORD— optional, gates all dashboard pagesAPPLICATIONINSIGHTS_CONNECTION_STRING— optional, enables telemetry
Run with Docker
# Build the production image
docker build -t copilot-insights ./app
# Run it against your PostgreSQL instance
docker run -p 3000:3000 \
-e DATABASE_URL="postgresql://user:pass@host:5432/copilot_insights?sslmode=require" \
-e ADMIN_PASSWORD="<strong-password>" \
copilot-insightsDatabase migrations run automatically on startup. Once the container is up, open http://localhost:3000 (or your proxied URL), go to Settings, and configure your GitHub token and sync schedule.
Tip
For local development, use the Quick Start above (pnpm run dev) instead of building a container.
ghcp-dashboard/
├── app/ # Next.js application
│ ├── src/
│ │ ├── app/ # Pages and API routes
│ │ │ ├── api/ # REST API endpoints (31 routes)
│ │ │ ├── metrics/ # Copilot Usage dashboard
│ │ │ ├── code-generation/ # Code generation report
│ │ │ ├── pull-requests/ # PR & Autofix report
│ │ │ ├── agents/ # Agent impact report
│ │ │ ├── ai-adoption/ # AI adoption cohorts report
│ │ │ ├── cli/ # CLI impact report
│ │ │ ├── seats/ # Licensing page (live)
│ │ │ ├── ai-credits/ # AI credits usage + billing dashboard
│ │ │ ├── premium-requests/ # Historical premium requests report
│ │ │ ├── models/ # AI model catalog
│ │ │ ├── users/ # User explorer
│ │ │ ├── enterprise-teams/ # Enterprise team management
│ │ │ ├── reference/ # Metrics reference
│ │ │ └── settings/ # Configuration, data sync, audit log & app info
│ │ ├── components/ # Shared React components
│ │ ├── lib/ # Database, ETL, i18n, theme, utilities
│ │ └── types/ # TypeScript type definitions
│ ├── drizzle/ # Database migrations
│ ├── public/ # Static assets
│ ├── Dockerfile # Multi-stage production build
│ └── package.json
├── infra/ # Azure Bicep IaC
├── docs/ # Documentation & screenshots
└── azure.yaml # Azure Developer CLI config
pnpm run dev # Start dev server
pnpm run build # Production build
pnpm run start # Start production server
pnpm run lint # ESLint check
pnpm run db:generate # Generate Drizzle migrations
pnpm run db:migrate # Run migrations
pnpm run db:push # Push schema to DB
pnpm run ingest # Manual data ingestThe dashboard supports two sync modes:
- Auto-sync: Background sync on a configurable interval (default: every 6 hours). Runs via
instrumentation.tson server startup. - Manual sync: Trigger from the Settings → Data Sync page via SSE streaming.
Both modes call the GitHub Copilot Usage Metrics API (v2026-03-10) and transform the data into a star schema for analytics queries.