Open-Source Web Intelligence

The Search & Web Retrieval Layer Built for AI Agents

Searqon searches, crawls, extracts, and purifies raw web pages into clean markdown. Self-hosted, privacy-first, and highly performant.

searqon-server logs
sooriya@searqon:~$ |
2026/06/16 07:58:44 [Search] Query="deepseek models" Limit=3 Scrape=true
2026/06/16 07:58:47 [Search] Provider=searxng found 3 results
2026/06/16 07:58:47 [Scrape] Match agent policy for "https://deepseek.com" -> Allowed
2026/06/16 07:58:47 [Scrape] Attempting Lightpanda fetch: https://deepseek.com
2026/06/16 07:58:49 [Lightpanda] Scraped page success (824 words, 2055ms)
# Markdown Response # DeepSeek DeepSeek is an artificial intelligence company specializing in large-scale language models. ## Open Source Models - DeepSeek-V3: Mixture-of-Experts language model - DeepSeek-Coder: Specialized software development agent

Dual-Provider Search

Queries a local SearXNG cluster as the primary engine, falling back gracefully to DuckDuckGo Lite HTML if SearXNG is rate-limited or down.

Sub-5ms Cache Check

Maintains a dual-layer PostgreSQL cache pool. Search results (24h TTL) and page scrapes (7-day TTL) load instantly for recurring requests.

Lightpanda Rendering

Spawns the Lightpanda headless browser dynamically to render client-side JavaScript (React, Next.js) for deep text extraction.

Robots.txt Obedience

Audits remote rules, respects crawling limits, respects crawl-delays, and gracefully switches User-Agents according to site policies.

SearqonBot Verification

SearqonBot/1.0 is the web crawler utilized by self-hosted Searqon search installations. It compiles public web content to provide structured text data for users and AI summarization apps.

We value webmaster policies and work hard to ensure the bot behaves politely:

Anti-Overload Safeguards

Honors crawl delays specified in robots.txt and restricts timeouts to 7-8 seconds to avoid pinning server threads.

Opt-Out Capabilities

Allows webmasters to completely block or route SearqonBot using standard instructions in their robots.txt file.

Zero Security Intrusion

We only scrape public webpages. We never scan paths for vulnerabilities or bypass security payloads.

Official User-Agent string
SearqonBot/1.0 (+https://sooriya04.github.io/Searqon/)

                    
Text copied to clipboard!