Climate spending intelligence for Canadian public funding. Track where ~$321B in green money goes, who lobbies for it, and what patterns emerge. Figures are approximate and as of the last data refresh.
Canada spends an estimated ~$321B through climate-related departments, and roughly 70,000 lobbying registrations target these departments. These public datasets are normally siloed, and few tools connect them.
CarbonTrail cross-references procurement, grants, and lobbying data to make climate spending transparent. We are pro-climate — transparent spending builds public trust. Public trust protects climate budgets.
| Feature | Description |
|---|---|
| Dashboard | Live stats, top signals, spending timeline (Recharts) |
| Green Money Flow | Sankey diagram: Department → Program → Recipient |
| Anomaly Detection | Spending spikes, sole-source concentration, new vendor signals |
| Funding Gaps | Interactive Canada map, provincial equity analysis |
| Feature | Description |
|---|---|
| Lobby ↔ Funding Loops | Orgs that lobby AND receive climate $ (2,262 detected) |
| Greenwash Radar | Energy-tax nexus, broad spectrum lobbying patterns |
| Network Graph | Force-directed visualization of org ↔ department connections |
| Entity Profiles | Full dossier: grants, contracts, lobbying, timeline, AI analysis |
| Compare | Side-by-side organization comparison |
| Feature | Description |
|---|---|
| AI Chat | Conversational data exploration (Bedrock/Claude + fallback) |
| AI Explanations | One-click pattern analysis on every entity |
| Climate Programs | Deep-dive into 100+ individual programs with charts |
| Feature | Description |
|---|---|
| ⌘K Command Palette | Instant search across entities + page navigation |
| Watchlist | Track organizations (localStorage) |
| CSV Export | Download any table as CSV |
| Print Reports | Print-friendly entity profiles |
| Keyboard Shortcuts | Press ? for shortcut panel |
| Responsive | Mobile hamburger menu, auto-collapse sidebar |
| Toast Notifications | Feedback on watchlist, export, share actions |
| Source | Records | Coverage |
|---|---|---|
| PSPC Contracts | 1,261,485 | Federal procurement (258K climate-relevant) |
| Federal Grants | 1,276,108 | Grants & contributions (113K climate-relevant) |
| Lobbyist Registry | 127,415 | All registrations (70K target climate depts) |
| Lobbyist Subjects | 996,271 | Subject matter codes |
| T3010 Charities | 83,581 | Registered charities |
| Charity Directors | 568,144 | Board members |
Total: 5.2M+ records analyzed, 747 MB DuckDB database
Frontend: Next.js 16 · TypeScript · Tailwind CSS · shadcn/ui · Recharts · d3-sankey · react-force-graph-2d
Backend: Python FastAPI · DuckDB (embedded OLAP)
AI: AWS Bedrock / Claude (with rule-based fallback)
Data: Canadian Open Data (Open Government Licence)
# 1. Clone
git clone https://github.com/ctmakc/carbontrail
cd carbontrail
# 2. ETL (one-time — downloads nothing, uses local CSVs)
cd backend
pip install -r requirements.txt
python etl/ingest.py # ~67 seconds, creates 747MB DuckDB
# 3. Backend
uvicorn app.main:app --port 8902
# 4. Frontend (new terminal)
cd frontend
npm install
npm run dev # http://localhost:3000┌────────────────────┐ ┌───────────────────────┐
│ Next.js 16 (18pg) │────▶│ FastAPI (35 endpts) │
│ Tailwind + shadcn │ │ 13 routers │
│ Recharts · d3 │ │ AI service (Bedrock) │
│ react-force-graph │ └───────────┬────────────┘
└────────────────────┘ │
┌────────▼────────┐
│ DuckDB (747MB) │
│ 17 tables │
│ 5.2M records │
└─────────────────┘
All figures are approximate and as of the last data refresh. They are derived from public records and update whenever the ETL pipeline re-runs. Every item below is a review signal — a pattern that may warrant a closer look — not an allegation or a finding of wrongdoing. Lobbying is legal and often legitimate, and receiving public funding is not evidence of anything improper. Name cross-matching and keyword tagging are heuristic and can produce false positives, so each signal requires independent verification against the source records.
- ~$321B tracked in climate-related public spending
- ~2,262 lobby-to-funding loops detected (org-name cross-match; may include false positives)
- ~3,559 dual recipients (grants + contracts from same departments)
- ~$8.6B in contracts recorded as 100% sole-source (e.g. Chantier Davie) — a procurement pattern to review, not an irregularity in itself
- ~2,776 greenwash review signals (anomalous lobbying-breadth patterns)
- NextStar Energy: ~$14.5B grant alongside 6 lobbying registrations — a pattern shown for review, drawn from public records
🌱 Pro-climate — transparency protects green budgets 🔍 Review signals, not allegations — careful language always 📎 Source-linked — every finding traces to public records ⚖️ Responsible — "pattern", "signal", "requires verification"
MIT
Built with conviction that climate action works better in the light. 🌍