Everything below is published on npm, documented, and running in production somewhere. Two of them you can run on this page, right now.
Universal document viewer for the web
One component, any file. Renders PDFs, spreadsheets, presentations, code, media, and more — with full-text search, OCR, zoom, and dark mode. 14+ formats, tree-shakeable, framework-agnostic.
npm install doclensPDFium-powered PDF library for JavaScript
Chrome-grade PDF rendering via WebAssembly with full TypeScript types. Text extraction, search, bookmarks, annotations, forms, and digital signatures. Powers doclens under the hood.
npm install pdfnovaAI-powered test generation that actually works
Generates, runs, and auto-fixes tests in a loop until they pass. AST analysis via the TypeScript compiler API, coverage-gap filling from lcov/cobertura, works with any LLM.
npm install -D @amit641/testpilot-aiLightweight universal AI client for JS & TS
Zero dependencies. Works everywhere — browser, Node, Deno, edge. Unified API for OpenAI, Anthropic, Google, and any OpenAI-compatible provider. Streaming, tool calls, structured output.
npm install aiclientjsDrop-in observability and cost tracking for any LLM SDK
One line of code to track spend in USD, token counts, latency (TTFT), and error rates for any LLM SDK. Costs computed locally from a versioned price table, written to a sink you control.
npm install @amit641/llmmeterThis is my resume, rendered live in your browser by doclens + pdfnova — the exact packages above, installed from npm. PDFium compiled to WebAssembly, running right here.
Switch the provider. The code doesn’t change — that’s aiclientjs. Zero dependencies, runs in browser, Node, Deno, and edge runtimes.
import { createClient } from "aiclientjs";
const ai = createClient({
provider: "openai",
model: "gpt-4o",
});
// Same streaming API for every provider — no SDK juggling.
for await (const chunk of ai.stream({
messages: [{ role: "user", content: "Explain microfrontends" }],
})) {
process.stdout.write(chunk.text);
}