parallel-web

Search the web, extract URL content, and run deep research via the Parallel Web Systems APIs.

Updated Oct 7, 2022
One-click install
npx skills add https://github.com/tamagusko/linux-cfg --skill parallel-web-tamagusko
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: parallel-web
Source: https://github.com/tamagusko/linux-cfg/tree/main/dotfiles/claude/skills/parallel-web
Command: npx skills add https://github.com/tamagusko/linux-cfg --skill parallel-web-tamagusko

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openai, parallel-web, and includes scripts (resource) and references (resource) components.

What problem does it solve? Gathering current web information, verifying citations, and producing multi-source research reports normally requires juggling search engines, manual page reading, and note-taking. This Skill consolidates web search, URL extraction, and deep research into one CLI and Python interface backed by the Parallel Web Systems APIs, returning synthesized summaries with citations. ## Core Features & Use Cases - Web Search: Query the Parallel Chat API (base model) to get synthesized answers with inline citations and a deduplicated source list. - Deep Research: Run comprehensive multi-source research reports (core model) for market analysis, literature background, or competitive intelligence. - URL Extraction: Extract clean markdown from specific URLs, including PDFs and JavaScript-heavy pages, for citation verification. - Use Case: While writing a scientific paper section, run a deep research query on the topic, save the cited report to a sources folder, then verify each citation's metadata by extracting its DOI URL. ## Quick Start Set the PARALLEL_API_KEY environment variable, then ask the assistant to search the web for a topic and save the cited results into the project's sources folder.

Frequently Asked Questions about parallel-web

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a web search with the Parallel API from Python?

Use the ParallelSearch class or the CLI search subcommand with a natural language objective. The Chat API base model returns a synthesized summary with inline citations and a deduplicated sources list.

What is the difference between the base and core chat models?

The base model handles standard research and factual queries with 15-100 second latency, while the core model performs complex multi-source synthesis taking 1-5 minutes. Search defaults to base and deep research defaults to core.

When should I use the Extract API instead of search?

Use Extract only when you already have a specific URL, such as verifying a citation or reading a known documentation page. For finding information broadly, use search or deep research instead.

Does the Parallel Extract API work with PDFs and JavaScript pages?

Yes, the Extract API renders JavaScript-heavy pages and converts PDFs to clean markdown automatically. You can request focused excerpts aligned to an objective or the full page content.

Why do I get a PARALLEL_API_KEY not set error?

The client requires the PARALLEL_API_KEY environment variable before any API call. Export it in your shell after obtaining a key from platform.parallel.ai, then rerun the command.

What are the limitations of the deep research command?

Deep research queries are limited to 15,000 characters and can take 1-5 minutes with the core model. Vague queries produce unfocused results, so include specific aspects, time bounds, and output expectations.