The AI research assistant that cites real sources honestly
Web Researcher MCP is an open-source AI citation verification tool that connects large language models to research capabilities designed around provenance rather than mere relevance. It serves developers, legal professionals, academics, and anyone who relies on AI-generated research that must be factually grounded. The core value is restoring trust in AI-assisted research by providing tools that verify every source, detect self-promotion in recommendations, and enable search restricted to authoritative domains. Unlike standard web search, this tool labels each source with domain reputation and authority tier, so the AI assistant can distinguish an independent review from a company's own marketing page.
The tool directly addresses the epidemic of sloptimization — the systematic manipulation of web content to train chatbots into recommending brands — and the parallel crisis of AI citation rot. Chatbots cannot distinguish between semantic relevance and actual authority; they treat a company's self-promotional listicle the same as an independent analysis. Over 60% of news articles now contain miscited sources, and fabricated citations have appeared in legal filings and academic papers. Web Researcher MCP solves the core failure: chatbots lack the tools to check whether a source exists, whether it supports the claim it is cited for, and whether the recommender has a conflict of interest. By giving AI assistants these verification capabilities, the tool restores the standard that careful human researchers have always applied.
The verify_recommendation tool is the first line of defense against sloptimization. When an AI assistant recommends a product or service, verify_recommendation runs four parallel checks on each recommended item. First, it detects self-promotion by fetching the source URL and checking whether that domain's own brand appears at position one in the ranking — a lexical flag that catches company blogs ranking themselves first. Second, it checks for conflicts of interest by examining author bios against the recommended entity. Third, it evaluates domain reputation using a curated dataset that assigns authority tiers. Fourth, it verifies link liveness, flagging dead URLs with their HTTP status codes. The output is an evidence array of flags, not a verdict; the AI assistant can then decide how to present the information. This feature directly counters the most common sloptimization tactic: branded listicles that rank the company number one.
The verify_citation tool addresses the epidemic of AI citations that point nowhere or to retracted work. Given a DOI, URL, or free-text reference, it runs three checks in parallel against authoritative registries including Crossref, OpenAlex, and the doi.org handle registry. It distinguishes between a source that is not found (possible fabrication) and one that simply could not be checked (like a book). It queries Crossref Retraction Watch to identify retracted papers. It also checks whether the link still resolves, and if not, automatically queries the Internet Archive for a Wayback snapshot. A patent innovation is the optional claim parameter: when provided, the tool fetches the source text and performs a lexical coverage check to see if the source actually addresses what the citation claims it proves. This catches mischaracterized sources — real papers cited for claims they do not support — which is the specific failure mode appearing in legal sanctions and academic retractions.
Publisher
A
admin
Launch Date2026-06-15
Pricingfree
Domain Authority
Ahrefs DR99/100