TypeScript

On the web: Hiki and Scout

Hiki and Scout are both Next.js 15 apps with React 19 and the App Router, written in TypeScript. Hiki’s front end renders a multilingual catalog with next-intl and locale-aware routing, and readers filter it by 21 genres, 100+ tags, price, discount, stock and store. Scout’s front end is where people run an AI sales agent: a seven-step setup wizard, a drag-and-drop Kanban pipeline, an analytics dashboard and an activity log that updates live while the agent works. Its agent is a modular pipeline with discrete steps to discover, enrich, qualify, expand and draft, and its data sits in SQLite in WAL mode.

Hiki is also where TypeScript stops. Its back end, which scrapes the bookstores and enriches every title, is Python with FastAPI.

On the desktop: Bridddge

Bridddge is a macOS app for LinkedIn, built with Electron, React and TypeScript. A multi-page Vite build gives each feature its own React app: the dashboard, the CRM, Social Selling Index analytics, the scheduler and the automations. Data lives in a local SQLite database, so nothing goes to the cloud. Bridddge ships as a signed and notarized macOS app, distributed with an Apple Developer ID.

On the server: seo-mcp

seo-mcp is an MCP server written in Node.js and TypeScript, with no framework dependencies. It runs on the user’s own machine and needs no API keys. AI editors such as Claude Desktop, Claude Code and Cursor call it to audit a page or crawl a whole site. Node.js worker threads scan pages in parallel, which keeps crawls of large sites fast. Each of its 28 rules returns structured findings ranked by severity, written for an LLM to read and explain. For live monitoring it streams changes from Firehose by Ahrefs over Server-Sent Events and filters them with Lucene queries. The code is open source on GitHub.