One server, six categories — web, social, news, academic, code, and documents. No API keys, no in-server LLM. Each tool returns markdown for humans and structured JSON for agent chaining.
Concurrent multi-engine web search. Scrapling-backed TLS fingerprinting bypasses Cloudflare; a failure on any single engine never blocks the others. Results are deduplicated and merged, each annotated with a quality score plus an aggregate confidence signal.
query, num_results, extract_content, follow_links, max_depth
Searches nine developer and community platforms in parallel, no authentication required. Ideal for surfacing real-world sentiment, bug reports, and discussion that never makes it into web search results.
query, platforms, max_results_per_platform, reddit_subreddit, time_filter
Five news sources merged and time-windowed, deduplicated by URL across engines. Pass a freshness filter to scope results to the last day, week, or month.
query, time_range (day · week · month · anytime)
Searches public GitHub repositories via the official API with built-in rate limiting (60/hr unauthenticated). Optionally pulls README content for deeper context.
query, language, sort (stars · forks · updated), include_readme, max_results
Structured site traversal in one of three modes. Returns per-page quality scores and an aggregate confidence signal so a caller can trust or discard the crawl as a whole.
url, mode, depth & page limits
One tool, six operations, dispatched by an operation enum. The Swiss-army knife of the toolset — turn any URL or text into clean, scored, or reconciled output. Returns markdown plus structured JSON.
find_conflicts surfaces numeric, date, and polarity disagreements across sources as a first-class signal — instead of silently averaging them away.
Two modes behind one tool. topic runs an open-ended search-and-rank; entity builds a unified cross-source profile of any named person, company, or product, fanning out across web, news, GitHub, social, and academic in parallel.
query, mode
Extracts text from documents and images — including OCR for scans and screenshots via EasyOCR, lazy-loaded with no setup. Handles files up to 50 MB.
Up to 50 MB per file · EasyOCR for image extraction
Academic paper search and dataset discovery in one call. Returns relevance-ranked, citation-scored results across five paper providers and four dataset hubs.
One URL, every MCP client. No API keys, no signup.