diff --git a/CLAUDE.md b/CLAUDE.md new file mode 100644 index 0000000..3fa2eee --- /dev/null +++ b/CLAUDE.md @@ -0,0 +1,258 @@ +# CLAUDE.md — DumpIt + +This file is the single source of truth for AI-assisted development on DumpIt. +Read this before writing any code, suggesting any feature, or making any architectural decision. + +--- + +## What DumpIt Is + +DumpIt is a personal knowledge vault with AI-powered retrieval. The core loop: + +1. User saves something (a link, a note, a PDF) +2. User asks a question in natural language +3. DumpIt returns a cited answer grounded in what the user actually saved — not hallucinated from general knowledge + +It is **not** a general-purpose chatbot. It is **not** a task manager. It is **not** an autonomous agent. +It is a save-and-query tool that happens to have excellent AI retrieval underneath. + +**One-line positioning:** "Save anything. Ask anything. Get answers from your own knowledge — not a stranger's AI." + +--- + +## Who It's For + +Primary persona: knowledge workers, developers, indie hackers, researchers, students — anyone who saves things online and later can't find or remember them. + +The Canva/Claude parallel is intentional: **simple enough for a non-technical user to start in 60 seconds, deep enough for a developer to plug into their own stack via MCP or API.** + +Two user modes exist: +- **Consumer mode** — one-click save via browser extension, ask in plain English, get cited answers. No setup. +- **Power-user mode** — MCP server, API access, export, query from Claude Desktop or Cursor. Full control. + +--- + +## Tech Stack + +| Layer | What | +|---|---| +| Framework | Next.js 14 (App Router) | +| Language | TypeScript | +| Styling | Tailwind CSS — **Zinc palette** (bg-zinc-50 / dark:bg-zinc-950). Do not switch to Slate or Gray. | +| Typography | `Space Grotesk` (headings, badges, UI labels) + `Inter` (body, inputs) via `next/font/google` | +| Components | shadcn/ui | +| Auth | Google Auth only — email/password was deliberately removed. Do not re-add it. | +| Database | Firebase (Firestore) + Firebase Admin SDK for server-side ops | +| AI/Embeddings | Gemini (embeddings + generation). Model: `gemini-2.5-pro` via env var `GEMINI_MODEL` | +| Rate limiting | Upstash Redis | +| Error tracking | Sentry (client + edge middleware + server handlers) | +| Email | Resend + React Email | +| Pagination | Cursor-based. Do not revert to offset-based. | +| Payments | Dodo Payments (Merchant of Record — handles VAT/GST/global tax automatically) | +| Deployment | Vercel (frontend), Firebase (backend/db) | + +--- + +## Environment Variables + +``` +# Firebase client (public) +NEXT_PUBLIC_FIREBASE_API_KEY= +NEXT_PUBLIC_FIREBASE_AUTH_DOMAIN= +NEXT_PUBLIC_FIREBASE_PROJECT_ID= +NEXT_PUBLIC_FIREBASE_STORAGE_BUCKET= +NEXT_PUBLIC_FIREBASE_MESSAGING_SENDER_ID= +NEXT_PUBLIC_FIREBASE_APP_ID= + +# Firebase admin (server only — never expose to browser) +FIREBASE_PROJECT_ID= +FIREBASE_CLIENT_EMAIL= +FIREBASE_PRIVATE_KEY= + +# Gemini +GEMINI_API_KEY= +GEMINI_MODEL=gemini-2.5-pro + +# Upstash Redis (rate limiting) +UPSTASH_REDIS_REST_URL= +UPSTASH_REDIS_REST_TOKEN= + +# Sentry +SENTRY_DSN= + +# Resend (email) +RESEND_API_KEY= +RESEND_FROM_EMAIL= + +# Dodo Payments +DODO_PAYMENTS_API_KEY= +DODO_WEBHOOK_SECRET= +``` + +--- + +## Repository Structure + +``` +app/ Next.js App Router pages and API routes + (auth)/ Auth-gated routes + api/ API route handlers (server-side only) + resources/ CRUD for saved items + search/ RAG query endpoint + digest/ Weekly email digest + u/[username]/ Public user profile pages +docs/ Internal documentation +public/ Static assets +types/ Shared TypeScript types +``` + +--- + +## Core Features — Current State + +| Feature | Status | +|---|---| +| Google Auth (sign in/out) | ✅ Live | +| Save links (URL fetch + metadata extraction) | ✅ Live | +| Save notes | ✅ Live | +| Tag organization | ✅ Live | +| Search & filter | ✅ Live | +| Public/private toggle per resource | ✅ Live | +| Public user profiles (`/u/[username]`) | ✅ Live | +| Social sharing (LinkedIn, Facebook) | ✅ Live | +| Edit resources | ✅ Live | +| AI-powered cited Q&A (RAG) | ✅ Live | +| Upstash rate limiting on AI routes | ✅ Live | +| Cursor-based pagination | ✅ Live | +| Weekly email digest (Resend) | ✅ Live | +| Sentry error monitoring | ✅ Live | +| Browser extension (one-click save) | 🔨 Building — highest priority | +| PDF upload + text extraction | 🔨 Next after extension | +| Email-to-save (`save@dumpit.page`) | 📋 Planned | +| Mobile share sheet | 📋 Planned | +| MCP server (query vault from Claude/Cursor) | 📋 Planned — key differentiator | +| Dodo Payments integration | 🔨 In progress | +| Tone/preference profile (personalized answers) | 📋 Planned — v2 | + +--- + +## V1 Scope — Do Not Expand Beyond This + +V1 is complete when: +1. Browser extension ships (one-click save from any tab) +2. PDF upload and text extraction works +3. Dodo Payments is live (founding member plan: $9/mo, 50% off) +4. RAG-cited Q&A is sharp and reliable on real saved content +5. Real users (not fake numbers) are using and paying + +**Do not add to v1 scope.** Every feature that isn't on this list is v2 or later. + +--- + +## What Has Been Explicitly Ruled Out + +These decisions were made deliberately. Do not reopen them without a strong reason. + +### Authentication +- Email/password auth was removed. Google Auth only. Rationale: simpler onboarding, zero password-reset overhead. + +### Autonomous agents / "act on my behalf" +- DumpIt answers questions. It does not send emails, book things, or take irreversible actions on behalf of the user. +- This is a v1 and v2 constraint. If this ever changes it must be a deliberate, separate decision — not a feature added incrementally. + +### Digital clone / impersonation +- DumpIt does not generate content that impersonates the user to other people. +- The tone/preference profile (v2) shapes *how DumpIt answers the user*, not how it represents the user externally. + +### OS-level integration +- Not a v1 or v2 item. The browser extension covers the majority of the capture use case with a fraction of the complexity. + +### Fake social proof +- The landing page must not show fabricated user counts, fake testimonial avatars, or invented counters. +- Only show real numbers. If a number is zero, either omit it or show zero honestly. + +### Offset-based pagination +- Already migrated to cursor-based. Do not revert. + +--- + +## Competitive Context + +| Competitor | What they are | Why DumpIt is different | +|---|---|---| +| Supermemory | Developer memory API/infrastructure | Dev-only, no consumer UX, requires engineering setup | +| Mem0 | Open-source memory layer for AI agents | Same — infrastructure, not a product | +| Hermes + OMI + Obsidian | Power-user second brain stack | 30-45 min technical setup, not consumer-friendly | +| Notion AI / Obsidian | Note-taking + AI | Organization-first, not retrieval-first; no capture layer | +| Orchid | Consumer AI assistant via text | Acting-on-your-behalf positioning, not knowledge retrieval; receiving backlash | +| Claude / ChatGPT memory | Native AI memory | No capture layer, no extension, no email-forward; setup-heavy for real knowledge base use | + +**The gap DumpIt fills:** retrieval-first, consumer-simple, with a developer escape hatch (MCP/API). Nobody has shipped this cleanly. + +**The risk:** platform memory features (Claude Projects, ChatGPT Memory) are improving fast. The window for "it just works out of the box" as a moat is real but not permanent. Ship fast, get real retention data, and decide whether this is a 2-3 year window product or something with a deeper moat. + +--- + +## Design System Rules + +- **Color:** Zinc palette only. `bg-zinc-50` light, `bg-zinc-950` dark. High-contrast borders: `zinc-200` / `zinc-800`. +- **Typography:** `Space Grotesk` for headings, badges, UI labels. `Inter` for body copy, inputs, descriptions. +- **Theme toggle:** MagicUI-style animated circle reveal transition. Already implemented — don't replace it. +- **Components:** shadcn/ui primitives. Don't introduce a second component library. +- **Motion:** Subtle. This is a productivity tool, not a marketing site. Animations should feel fast, not showy. + +--- + +## API Conventions + +All list responses follow this shape: +```typescript +{ + data: T[], + total: number, + nextCursor: string | null, // cursor-based pagination + hasMore: boolean +} +``` + +Server-side API routes use Firebase Admin SDK. Client-side code uses the Firebase client SDK for auth only — all database operations go through API routes, never direct client Firestore access. + +Rate-limited routes (anything calling Gemini) must check Upstash Redis before making the AI call. Pattern is in the existing `/api/search` route — follow it. + +--- + +## Pricing + +| Plan | Price | Notes | +|---|---|---| +| Free | $0 | Limited saves, limited queries/month | +| Founding Member | $9/mo | 50% off forever, first 1000 members only | +| Regular | $18/mo | After founding member slots fill | + +Payment processor: Dodo Payments (MoR — handles all global tax compliance automatically). + +--- + +## Founder Context (for tone and decision-making) + +- Solo founder: Rayan, Full Stack + UI/UX + blockchain developer, Forward Deployed Engineer at build.ai, Bangalore. +- Goal: ship something real, get paying users, build a credible product story (for YC application pipeline and potential acquisition). +- North star for this product: **real users, real retention, real revenue** — not metrics theater. +- Open source: the codebase is MIT licensed and public. This is intentional. +- Target market: US-facing, global reach. + +--- + +## What Good Looks Like + +A session is successful if: +- The core save→ask→cited-answer loop works reliably on real saved content +- A new user can be saving and querying within 3 minutes of signing up +- No feature outside v1 scope was added +- No fake numbers appear anywhere in the product or on the landing page +- Code follows the existing zinc/Space Grotesk design system +- All Gemini calls go through rate-limited API routes, never client-side + +--- + +*Last updated from claude.ai conversation context — August 2026.* diff --git a/app/components/landing/ApiSection.tsx b/app/components/landing/ApiSection.tsx index e31a24e..622f3e8 100644 --- a/app/components/landing/ApiSection.tsx +++ b/app/components/landing/ApiSection.tsx @@ -23,15 +23,15 @@ curl "https://app.dumpit.app/api/ai/search?q=RAG+citation+patterns" \\ const ApiSection = () => { return ( -
+
{/* Left */}
API & MCP -

+

Build on your vault.

-

+

DumpIt exposes a REST API and a Claude MCP integration so you can query your second brain programmatically — from your own tools, scripts, or AI agents.

    @@ -41,14 +41,14 @@ const ApiSection = () => { { title: 'Webhooks (roadmap)', body: 'Get notified when indexing completes or a resource is added by the extension.' }, ].map((item) => (
  • -
    {item.title}
    -
    {item.body}
    +
    {item.title}
    +
    {item.body}
    • ))}
API included in Pro @@ -56,16 +56,16 @@ const ApiSection = () => { {/* Right — code block */}
-
-
+
+
- DumpIt API + MCP + DumpIt API + MCP
- 
+            
               {codeSnippet}
diff --git a/app/components/landing/CTA.tsx b/app/components/landing/CTA.tsx index 65792e1..0f15118 100644 --- a/app/components/landing/CTA.tsx +++ b/app/components/landing/CTA.tsx @@ -9,22 +9,22 @@ const CTA = () => { const { user } = useAuth() return ( -
+
- + Shared knowledge

Build your own vault, then learn from what others share.

-

+

Keep personal resources private, publish useful links when they are worth sharing, and query both surfaces from one AI workspace.

{open === index && ( -
+
{faq.a}
)} diff --git a/app/components/landing/Features.tsx b/app/components/landing/Features.tsx index d428390..ed39c55 100644 --- a/app/components/landing/Features.tsx +++ b/app/components/landing/Features.tsx @@ -3,7 +3,7 @@ import { Chrome, Clock, FileText, Link2, StickyNote, Youtube } from 'lucide-reac const types = [ { icon: Link2, - label: 'Web links', + label: 'Links', description: 'Articles, docs, videos, tools — any URL. Text is fetched and indexed server-side.', status: 'live', color: 'text-blue-600 dark:text-blue-400', @@ -46,27 +46,27 @@ const types = [ label: 'More coming', description: 'Tweets, GitHub repos, Notion exports, and more on the roadmap. Shape it with your feedback.', status: 'roadmap', - color: 'text-slate-500 dark:text-slate-400', - bg: 'bg-slate-100 dark:bg-slate-800', + color: 'text-zinc-500 dark:text-zinc-400', + bg: 'bg-zinc-100 dark:bg-zinc-800', }, ] const statusLabel: Record = { live: { label: 'Live', style: 'bg-emerald-100 text-emerald-700 dark:bg-emerald-950/60 dark:text-emerald-400' }, 'coming-soon': { label: 'Coming soon', style: 'bg-blue-100 text-blue-700 dark:bg-blue-950/60 dark:text-blue-400' }, - roadmap: { label: 'Roadmap', style: 'bg-slate-100 text-slate-500 dark:bg-slate-800 dark:text-slate-400' }, + roadmap: { label: 'Roadmap', style: 'bg-zinc-100 text-zinc-500 dark:bg-zinc-800 dark:text-zinc-400' }, } const ContentTypes = () => { return ( -
+
Content types -

+

Not just links.

-

+

DumpIt is your AI second brain for everything you consume — web, notes, PDFs, and more.

@@ -85,8 +85,8 @@ const ContentTypes = () => { {badge.label}
-

{type.label}

-

{type.description}

+

{type.label}

+

{type.description}

) })} diff --git a/app/components/landing/FinalCTA.tsx b/app/components/landing/FinalCTA.tsx index 0a41418..3a0450e 100644 --- a/app/components/landing/FinalCTA.tsx +++ b/app/components/landing/FinalCTA.tsx @@ -9,13 +9,13 @@ const FinalCTA = () => { const router = useRouter() return ( -
+
{/* Background glow */}
- + Limited founding member offer @@ -25,14 +25,14 @@ const FinalCTA = () => {
you learn. -

- Every link you save and forget is knowledge lost. DumpIt makes your saved content answerable. +

+ Everything you save and forget is knowledge lost. DumpIt makes what you save answerable.

{user ? (