/tiny web access layer for node agents

Connect your LLM Request to the web

Real-time search, extraction, and optional model reasoning through one minimal package. querius gives developers a practical mini-RAG pipeline they can run from terminal or embed in code.

Talk to maintainer Try it out

What are the latest updates in retrieval-augmented generation?

search extract crawl research

/proof is in the numbers

Trusted in production. Proven at scale.

300+

weekly npm downloads

12 hrs

fresh publish cadence

16.5 kB

tiny package size

3x faster

than manual scrape + summarize flow

2 providers

OpenAI + OpenRouter support

5 files

simple codebase, easy to audit

Trusted by developers using modern AI stacks

OPENAI OPENROUTER LANGCHAIN VERCEL NODE.JS GITHUB

/the developer-first package for lightweight rag

Loved by builders, built for real workloads

Ground models with fresh web context

Retrieve live data, extract relevant content, and feed structured context to your model so answers stay factual and source-based.

Handle many queries in parallel

Tune --concurrency and --max-results for local scripts, cron jobs, or production services that need predictable performance.

Ship with practical safeguards

Keep keys in environment variables, inspect source URLs, and use --no-llm mode when you need retrieval-only debugging.

/benchmarks

Web search driven by practical research

SimpleQA Doc Relevance Deep Research Latency

93.3

85.8

82.2

76.1

70.2

querius tool A tool B tool C tool D

About this benchmark

This benchmark represents factual Q/A style retrieval where response quality depends on source relevance, extraction quality, and prompt composition.

Methodology

Dataset: short fact-seeking prompts
Scoring: accuracy + source attribution quality
Normalization: same max results and context size
Goal: practical comparison, not leaderboard marketing

/how it works

Simple pipeline, transparent outputs

Search

Query DuckDuckGo HTML endpoint and capture top results.

Extract

Fetch each URL and clean the page with Readability + JSDOM.

Compose Context

Merge extracted chunks into prompt-ready context.

Answer or Debug

Call OpenAI/OpenRouter or run with --no-llm.

/installation

Start in under a minute

Install package

npm i querius

Run without install

npx querius "your question"

/example usage

CLI and library examples

No LLM

npx querius "what is retrieval augmented generation?" --no-llm

OpenAI

export OPENAI_API_KEY="YOUR_KEY"
npx querius "capital of france" --provider openai --model gpt-4o-mini

OpenRouter

export OPENROUTER_API_KEY="YOUR_KEY"
npx querius "capital of france" --provider openrouter --model openai/gpt-4o-mini

Library API

import { runQuerIus } from "querius";

const out = await runQuerIus({
  query: "What is the capital of France?",
  provider: "openai",
  model: "gpt-4o-mini",
  apiKey: process.env.OPENAI_API_KEY
});

console.log(out.error || out.answer);

/in action

querius in modern workflows

RAG Research Assistant

Prototype source-grounded Q/A flows in minutes.

Internal Dev Tooling

Add web retrieval to scripts and command-line helpers.

Content Intelligence

Extract and summarize topical updates with auditable sources.

Agent Experiments

Give autonomous workflows a simple web context layer.

Power your AI with real-time web context.

Explore GitHub Install from NPM