A nonpartisan, AI-powered tool that tracks, analyzes, and rates 2026 US congressional and gubernatorial candidates on their climate and environmental records — and their positions on AI governance and data center regulation.
-blueviolet){kind=icon}
The Climate Candidate Tracker gives voters, journalists, and researchers a clear, evidence-based picture of where 2026 candidates stand on two independent policy dimensions:
🌿 Climate Score — alignment with climate science on energy, land, transportation, pollution, resilience, and federal climate policy.
⚡ AI Policy Score — positions on AI governance and data center regulation: renewable energy requirements for data centers, water usage limits, AI safety oversight, algorithmic accountability, election integrity, and worker protections.
Both scores are fully independent. A candidate's climate record does not affect their AI policy score, and vice versa. A high fossil-fuel senator could still earn a high AI safety score; a green champion might oppose data center regulation.
Elections covered:
- 2026 US Senate primaries and midterm elections (competitive and key races)
- 2026 gubernatorial races (states with significant climate or AI policy relevance)
- 2026 US House battleground districts
Currently tracking 542 candidates across Senate, Governor, and House races.
All analysis is nonpartisan. Scores reflect policy alignment — not party affiliation.
- AI-powered climate analysis — Claude (Anthropic) produces a 0–100 climate score with a full written breakdown
- Multi-dimensional scoring — Analysis spans 6 climate dimensions: energy, land, transportation, pollution, resilience, and policy commitment
- Senate vs. Governor analysis — Separate prompts tuned to federal vs. state-level climate powers
- Independent AI policy analysis — Completely separate scoring system; does not touch climate scores or data
- 7 scored dimensions with weighted aggregate score (see methodology below)
- Data center district mapping — Candidates in major data center states (Virginia, Texas, Iowa, etc.) are flagged with contextual analysis
- Big Tech PAC donation tracking — FEC-sourced donation data flags potential conflicts of interest on AI regulation votes
- AI bill co-sponsorship tracking — Congress.gov data shows which AI governance bills each candidate has sponsored
- Candidate comparison tool — Compare two candidates on any climate or AI policy dimension
- Dual leaderboards — Top 10 by climate score, AI policy score, or combined average
- Advanced filters — Filter by AI score range, data center district, Big Tech funding, or AI bill co-sponsorship
- Automated weekly sync — Candidate data syncs from Ballotpedia every Monday
- Fossil fuel donation tracking — Each candidate flagged with donation level (low / moderate / high / unknown)
- Congressional district maps — Interactive Leaflet maps showing US Census 119th Congress district boundaries
- Race competitiveness ratings — Cook Political Report and Sabato's Crystal Ball
- Zero-result search empty state — When a name search returns no results, users see a "No candidates found for [term]" prompt with a + Request be added button
- Inline request form — Expands in place with name pre-filled; fields for state, office, optional party, and optional source URL
- Duplicate detection — Before saving, the API checks the candidates table with a case-insensitive match; duplicates show a mini candidate card with Yes/No confirmation instead of a form error
- Rate limiting — Cookie-based limit of 5 submissions per session per 24 hours
- Search analytics — Every zero-result search is logged to
search_analytics(debounced, once per unique term); a separate log entry records when users click through to the request form
- Candidate Requests section — Appears above the candidate list; shows a pending count badge and a full table of pending requests
- Pending table columns — Name, State, Office, editable Party dropdown, Source URL link, Submitted date, Add Candidate button, Reject button
- Add Candidate — Inserts directly into the
candidatestable using request data; disabled until party is set (required by schema); marks requestapprovedand moves it to the archive instantly - Reject — Marks request
rejectedand moves it to the archive - Collapsible archive — Shows all approved and rejected requests with status badges and reviewed dates
| Score | Meaning |
|---|---|
| 0–20 | Climate denier or active fossil fuel champion |
| 21–39 | Weak record; opposes most climate action |
| 40–69 | Mixed or moderate record |
| 70–89 | Strong climate record with some gaps |
| 90–100 | Ambitious climate leader |
| Dimension | What it covers |
|---|---|
| Energy & Emissions | Clean energy support, fossil fuel opposition, grid modernization, carbon pricing |
| Land & Water | Public lands protection, water rights, deforestation, ocean and biodiversity policy |
| Transportation | EV adoption, public transit investment, aviation and shipping emissions |
| Pollution & Health | Air quality, chemical regulation, plastics policy, environmental justice |
| Climate Resilience | Flood and sea level preparedness, wildfire policy, drought and heat response, disaster funding |
| Policy Commitment | Paris Agreement support, IRA/GND positions, EPA authority, climate finance, NEPA |
Senate candidates are analyzed on their federal legislative record: votes, co-sponsorships, committee positions, and stances on federal climate bills and agency authority.
Governor candidates are analyzed on state-level climate powers: renewable energy standards, utility regulation, executive orders, land use authority, building codes, and climate resilience investment.
| Score | Meaning |
|---|---|
| 0–20 | Actively opposes AI regulation; no data center requirements |
| 21–39 | Weak or absent AI governance positions |
| 40–69 | Mixed positions or limited public record |
| 70–89 | Strong AI governance advocate with some gaps |
| 90–100 | Comprehensive AI oversight champion |
| Dimension | Weight | What it covers |
|---|---|---|
| Data Center Energy | 25% | Supports renewable energy requirements for AI data centers |
| AI Safety & Oversight | 25% | Supports federal AI safety testing and oversight frameworks |
| Algorithmic Accountability | 15% | Supports transparency and accountability for automated decision-making |
| Water Usage Policy | 10% | Supports water usage limits and disclosure for data centers |
| Grid Impact | 10% | Supports grid modernization to handle AI infrastructure load |
| AI in Elections | 10% | Supports disclosure requirements for AI-generated political content |
| AI Economic Policy | 5% | Supports worker protections and creator rights in the AI economy |
| Factor | Impact |
|---|---|
| High Big Tech PAC donations (>$10k) | −10 points |
| Moderate Big Tech PAC donations ($1k–$10k) | −5 points |
| Co-sponsored an AI safety bill | +10 points |
| Member of Senate AI Caucus | +5 points |
| Commerce Committee member | +5 points |
| Made AI regulation a campaign issue | +5 points |
AI policy scoring is architecturally independent of climate scoring:
- Separate prompt (
lib/prompts/aiPolicyPrompts.js) explicitly instructs the model to ignore climate positions - Separate storage tables (
ai_analyses,ai_analyses_lite) — no shared columns with climate analysis - Separate API route (
/api/analyze-ai) — no shared code with/api/analyze - Separate score badge (⚡ blue) displayed alongside but independent of climate badge (🌿 green)
| Source | Use |
|---|---|
| Ballotpedia | Candidate data, race information, weekly automated sync |
| Wikipedia | Candidate background, incumbency history, race context |
| FEC API | Big Tech PAC donation data for AI policy conflict-of-interest flags |
| Congress.gov API | AI bill co-sponsorship tracking (DEFIANCE Act, NO FAKES Act, etc.) |
| Data Center District Map | Hardcoded map of states/districts with significant data center presence (lib/data/dataCenterDistricts.js) |
| OpenStates API | State legislature voting records (planned) |
| Cook Political Report | Race competitiveness ratings |
| Sabato's Crystal Ball | Race competitiveness ratings |
| Anthropic Claude API | AI analysis engine (Opus for Tier 1, Haiku for Tier 2) |
| US Census TIGERweb API | 119th Congress congressional district boundaries for interactive maps |
| Layer | Technology | Purpose |
|---|---|---|
| Framework | Next.js 15 (Pages Router) | Server-side rendering, API routes |
| UI | React 18 | Component-based interface |
| Styling | CSS custom properties | Dark/light theming, responsive layout |
| Database | Supabase (PostgreSQL) | Candidate storage, persistent analysis, caching |
| AI | Anthropic Claude API | Climate and AI policy scoring and analysis |
| Maps | Leaflet.js | Interactive congressional district and state boundary maps |
| Scraping | Cheerio + node-fetch | Ballotpedia HTML scraping |
| Hosting | Vercel | Deployment, serverless functions |
| Automation | Vercel Cron | Weekly Ballotpedia sync (Mondays 6am UTC) |
git clone https://github.com/Sparkleleuk/ClimateTracker.git
cd ClimateTracker
npm install
cp .env.example .env.localEdit .env.local with your credentials:
| Variable | Where to find it | Required |
|---|---|---|
ANTHROPIC_API_KEY |
console.anthropic.com → API Keys | ✅ Required |
SUPABASE_URL |
Supabase → your project → Settings → API | ✅ Required |
SUPABASE_SERVICE_ROLE_KEY |
Supabase → your project → Settings → API → Secret key | ✅ Required |
SYNC_SECRET |
Any long random string — e.g. openssl rand -hex 32 |
✅ Required |
FEC_API_KEY |
api.open.fec.gov | Optional — enables Big Tech PAC data |
CONGRESS_API_KEY |
api.congress.gov | Optional — enables AI bill tracking |
- Open your Supabase project → SQL Editor
- Paste and run the full contents of
supabase/schema.sql - Run migrations in order:
supabase/migrations/005_ai_policy_scoring.sql supabase/migrations/006_candidate_requests.sql - Run the seed script to populate all candidates:
node scripts/seed.mjs- (Optional) Run AI policy analysis on Senate/Governor candidates:
node scripts/runAIPolicyAnalysis.mjsnpm run devOpen http://localhost:3000.
The app deploys automatically to Vercel on every push to main.
Add these in Vercel → your project → Settings → Environment Variables:
ANTHROPIC_API_KEYSUPABASE_URLSUPABASE_SERVICE_ROLE_KEYSYNC_SECRETFEC_API_KEY(optional)CONGRESS_API_KEY(optional)
Visit /?admin=true in the browser and click the Sync button. You will be prompted for your SYNC_SECRET value.
ClimateTracker/
├── components/
│ ├── ClimateTracker.jsx # Main UI — candidate list, dual scoring, leaderboards, modals
│ └── DistrictMap.jsx # Leaflet district/state boundary map modal
├── pages/
│ ├── index.js # Entry point — loads candidates via getServerSideProps
│ └── api/
│ ├── analyze.js # Climate analysis endpoint
│ ├── analyze-ai.js # AI policy analysis endpoint (independent)
│ ├── compare.js # Climate comparison endpoint
│ ├── compare-ai.js # AI policy comparison endpoint (by dimension)
│ ├── candidates.js # Returns candidate list from Supabase
│ ├── sync.js # Ballotpedia sync endpoint (cron + admin)
│ ├── request-candidate.js # Save user candidate requests; dupe check + rate limit
│ ├── log-search.js # Log zero-result searches to search_analytics
│ ├── admin-requests.js # Admin: fetch pending + archived candidate requests
│ └── admin-request-action.js # Admin: approve (insert candidate) or reject a request
├── lib/
│ ├── constants/
│ │ ├── tiers.js # Analysis tier definitions
│ │ └── aiDimensions.js # AI policy dimension weights and rubric modifiers
│ ├── prompts/
│ │ ├── housePrompts.js # Tier 1/2 climate prompts for House candidates
│ │ └── aiPolicyPrompts.js # Tier 1/2 AI policy prompts (independent)
│ ├── data/
│ │ ├── candidatesFallback.js # Static fallback if Supabase is unavailable
│ │ ├── bigTechDonations.js # FEC API fetcher — Big Tech PAC donations
│ │ ├── aiBillsTracker.js # Congress.gov API fetcher — AI bill co-sponsors
│ │ └── dataCenterDistricts.js # Hardcoded data center district map
│ ├── analysis/
│ │ └── batchAnalyzer.js # Anthropic Batch API — climate and ai_policy types
│ ├── sync/
│ │ └── candidateSync.js # Sync logic — upsert, status detection, AI data refresh
│ ├── scoring/
│ │ └── algorithmicScore.js # Tier 3 rule-based scoring for House candidates
│ └── supabaseClient.js # Supabase client (server-side only)
├── services/
│ ├── climateApi.js # Client-side fetch wrapper for /api/analyze
│ └── compareApi.js # Client-side fetch wrapper for /api/compare
├── scripts/
│ ├── seed.mjs # Programmatic seed script — run once after schema setup
│ ├── scrapeHouseToSupabase.mjs # Run locally to bypass Ballotpedia WAF (AWS IPs blocked)
│ └── runAIPolicyAnalysis.mjs # Batch AI policy analysis for Senate/Gov candidates
├── supabase/
│ ├── schema.sql # Full database schema
│ └── migrations/
│ ├── 001_candidates.sql
│ ├── 002_filing_deadlines.sql
│ ├── 003_office_types.sql
│ ├── 004_house_candidates.sql
│ ├── 005_ai_policy_scoring.sql # ai_analyses, big_tech_donations, ai_bill_cosponsors
│ └── 006_candidate_requests.sql # candidate_requests, search_analytics
├── styles/
│ └── climate.css # Theme tokens, filter layout, responsive breakpoints
├── .env.example # Required environment variables (safe to commit)
└── vercel.json # Cron job configuration (Mondays 6am UTC)
The automated sync pipeline runs every Monday and:
- Scrapes Ballotpedia for Senate and Governor candidate pages
- Adds any new candidates not yet in the database
- Updates candidacy statuses as races evolve
- Refreshes Big Tech PAC donation data for Senate and incumbent House candidates
- Never deletes historical candidates or their stored analysis
| Feature | Status | Priority |
|---|---|---|
| US Senate candidates | ✅ Complete | — |
| Gubernatorial candidates | ✅ Complete | — |
| Supabase persistence | ✅ Complete | — |
| Ballotpedia auto-sync | ✅ Complete | — |
| Candidate comparison tool | ✅ Complete | — |
| US House candidates (battleground districts) | ✅ Complete | — |
| Congressional district maps (Census TIGERweb) | ✅ Complete | — |
| AI Policy Scoring System | ✅ Complete | — |
| FEC Big Tech PAC donation integration | ✅ Complete | — |
| Congress.gov AI bill co-sponsorship tracking | ✅ Complete | — |
| Data center district mapping | ✅ Complete | — |
| Candidate request flow (user submissions) | ✅ Complete | — |
| Search analytics (zero-result logging) | ✅ Complete | — |
| Admin candidate request review panel | ✅ Complete | — |
| FEC fossil fuel donation data (full integration) | 🔲 Planned | High |
| State legislature candidates | 🔲 Planned | Medium |
| Per-issue scoring breakdown | 🔲 Planned | Medium |
| Email alerts for score changes | 🔲 Planned | Low |
| Public API for researchers | 🔲 Planned | Low |
| Mobile app | 🔲 Planned | Low |
- Nonpartisan — Both scores reflect policy alignment only. Party affiliation is not a scoring factor.
- AI may contain errors — Analysis is AI-generated and may mischaracterize a candidate's record. Always verify with primary sources before publishing.
- Scores are independent — A candidate's climate score has no bearing on their AI policy score and vice versa. The scoring systems are architecturally separated.
- Positions change — Candidates update their platforms. Check the analysis date on each card and re-analyze if needed.
- Not a voting guide — This tracker covers climate and AI policy positions only. It does not evaluate candidates on any other issue or their overall fitness for office.
MIT License — see LICENSE for details.
- Anthropic for the Claude API
- Ballotpedia for open candidate data
- Supabase for database infrastructure
- US Census Bureau for TIGERweb district boundary API
- Cook Political Report and Sabato's Crystal Ball for race ratings