What changed at GotNerfed
Our own changelog. We catalog every silent vendor change — and we publish our own changes the same way. New features, breaking API changes, infrastructure shifts. RSS coming when we exit dev mode.
- feature
Subsidy Risk Index, Idle Burn category, Cmd+K, status page, browser extension landing
Twitter and Reddit research streams surfaced two contrarian moats nobody else has: Subsidy Risk Index predicts WHO will nerf next based on margin pressure (wrappers vs foundation models, runway stage, parent-company depth); Idle Burn names a brand-new pain category around credits draining during agent idle. Plus Cmd+K palette, a /status page, and a /extension landing page surfacing the Chrome MV3 install path.
- feature
Better-Auth + Stripe scaffold + magic-link signin
Email magic-link signin via Better-Auth, Drizzle schema for users/sessions/subscriptions/api-keys, Stripe Checkout + Customer Portal + webhook handler. Pricing page now switches CTAs to Stripe checkout when logged in. Activation requires Vercel Postgres / Neon provisioning + 4 Stripe products (see SETUP_STRIPE.md).
- feature
PWA + browser extension + auto-classify cron + comments + procurement PDF + sentiment
Manifest + service worker + web push notifications. Chrome MV3 extension v0 (chrome-extension/) injects badge on 18 vendor pricing pages. Scanner now fires AI Jury for every detected diff. Receipt comments via Upstash KV. Procurement-grade PDF report per vendor (print-styled). Reddit sentiment scrape via daily cron with sparkline chart.
- feature
Public API v1, Compare page, Forecasts leaderboard, AI Assistant chat, Cost Analyzer, Migration Assistant, Plan Optimizer
GET /api/v1/{receipts,vendors,...} with tiered rate limits and X-RateLimit headers. Compare 4 vendors side-by-side. Brier-score-tracked forecasts leaderboard. Claude Haiku 4.5 RAG-style assistant. Calculate at-risk dollars per AI subscription. Per-vendor migration playbooks with alternatives ranked by Nerf Index.
- feature
AI Jury (multi-LLM verdicts), Vendor Response Tracker, Predicted Impact, Forecaster, Vendor Risk heatmap
Three independent LLMs grade every receipt for harm with score + ruling + rationale. Vendor responses tracked with tone (apology/explanation/denial/deflection/silence). Heuristic impact estimator. Probabilistic 30/60/90-day forecast per vendor. Cross-vendor risk dashboard.
- feature
Initial v0 ship — Nerf Index, Live Ticker, Diff Viewer, Bluesky bot, daily scanner
Public receipts feed seeded with 7 verified events (Cursor June 2025, Claude Code rate limits, Anthropic Aug 2025 quality regression, OpenAI Codex April 2026 token-credits, Anthropic Pro removal, Copilot June 2026 usage-based shift). Per-event Nerf Score and per-vendor Nerf Index with public methodology. Bluesky bot auto-posts critical/major events. Daily Vercel cron scanner across 20 vendors.