ShipKit — Production Instance

• What it is: production-grade SaaS starter kit "ShipKit" — the user's most polished personal product. README in commercial-product style ("built by indie hackers who've shipped 4+ production SaaS apps with this exact stack"). 8 apps + 16 shared packages. (The folder name sahovat_shipkit is likely a fork/instance for Insurance group (UZ); the core is the generic ShipKit.)
• Type / status / role: web-app / SaaS boilerplate (monorepo) / active (last commit 2026-05-23) / solo (AI-assisted) — Davron 87+36=123 commits, Claude `<noreply@anthropic.com>` 41 (co-authored AI work). No outside contributors.
• Activity period: 2025-11-10 → 2026-05-23 (~6.5 months), 164 commits.

• Languages: TypeScript, Bun.
• Monorepo (Turborepo + Bun):
• apps/: web (landing/auth/pricing/blog), api (Elysia REST + OpenAPI + Scalar UI), admin (RBAC, user management), docs, e2e (Playwright), mcp (MCP server!), worker, warehouseapp.
• packages/ (16): ai, analytics, auth (Better Auth + RBAC), cache (Redis), cli, db (Drizzle/Postgres), email (React Email + Resend), entities, env, feature-flags, jobs, logger, storage, ui (shadcn + Tailwind v4), eslint/ts-config.
• Key features (README): Polar.sh payments (checkout, webhooks, subscriptions, monthly/annual), Better Auth RBAC + admin, AI-native (.claude/skills/, AGENTS.md, llms.txt, .codex, .cursor), dark mode, i18n (next-intl), SEO (sitemap/robots/JSON-LD/metadata), rate limiting + usage tracking, feature flags.
• Quality/infra: Playwright E2E (apps/e2e), lefthook (git hooks), GitHub Actions, docker-compose, OpenAPI+Scalar. The most "engineering-mature" of the personal projects (tests + CI + hooks + observability packages logger/analytics).

• User (123) + Claude (41) — effectively solo AI-assisted: built a complete SaaS starter kit.
• Polar payments layer (checkout/webhooks/subscriptions), auth/RBAC + admin, feature flags, jobs/worker, storage, analytics, logger, email (React Email/Resend).
• MCP server as its own app (apps/mcp) — a genuinely "AI-native" product (exposes functions via MCP).
• E2E tests (Playwright), git hooks (lefthook), CI (GitHub Actions) — engineering discipline.
• 2 large Claude Code sessions (9 MB + 12.6 MB jsonl) — substantial AI-assisted development.

Confirmed by README/structure.

• A full SaaS starter kit with monetization → out-of-the-box Polar.sh integration (checkout + webhooks + subscriptions) — solves the real "idea to paying customers" pain. Few boilerplates make it to payments-ready.
• AI-native project → .claude/skills/, AGENTS.md, llms.txt + own MCP server (apps/mcp) — the project is both developed with AI and exposes an AI interface. The strongest "modern, in-trend" signal (exactly the brand goal).
• Engineering maturity → Playwright E2E, lefthook, GitHub Actions CI, OpenAPI+Scalar, dedicated logger/analytics/feature-flags/jobs/storage/env packages — this is platform architecture, not a quick starter.
• E2E type safety (Elysia + Eden) + 16 reusable packages — codifying the user's "golden stack" into a sellable product.

• Sessions found: 2 in the local Claude Code sessions store for this project (9.0 MB + 12.6 MB jsonl — very large).
• What was done with AI: 41 commits with Claude as co-author — a significant share of development happened through Claude Code. The project is clearly designed for AI workflows (.claude/skills/, AGENTS.md, llms.txt, support for Cursor/Codex/Copilot).
• AI workflow patterns: a multi-agent environment (Claude + Codex + Cursor), custom skills, llms.txt for context — a textbook example of AI-assisted development, valuable as content about how the user works with AI.

• 8 apps + 16 packages in a single monorepo (solo + AI).
• Payments-ready (Polar), auth/RBAC, feature flags, MCP server, E2E, CI, observability.
• 164 commits over ~6.5 months; 2 large AI sessions (~21 MB of transcripts).
• Claimed: the stack on which "4+ production SaaS have been shipped."