# Journalistic Reasoning Prover MCP > Journalistic Reasoning Prover forces independent verification on content, tracing every factual claim back to a named primary source. It checks for false balance, detects invented sources (hallucinations), and mandates full conflict disclosure before publication. ## Overview - **Category:** productivity - **Price:** Free - **Tags:** journalism, fact-checking, ethics, verification, sourcing, attribution, reporting, media, prover, reasoning, hallucination, news ## Description AI agents can generate text that sounds authoritative but often fails basic journalistic standards. They cite 'experts' generally or invent studies—it’s fabrication, not sourcing. This MCP forces content to pass a rigorous ethical and factual audit. When you run the check, it doesn't just flag errors; it demands specific evidence: Who said it? What was the date? Where did they say it? And how was that data collected? If a claim can't meet these standards—like requiring independent corroboration from two separate sources—the process rejects it. It’s built around foundational principles like the SPJ Code of Ethics and Kovach & Rosenstiel’s verification model, making sure your content is factually defensible. You connect this MCP via Vinkius to any AI client, running a full audit every time you generate critical reporting. ## Tools ### validate_journalistic_reasoning Runs a full audit on content, checking for source fabrication, unverified claims, false balance, complete attribution, and concealed conflicts of interest. ## Prompt Examples **Prompt:** ``` Write an article about the new diet trend. According to experts, it helps you lose 10 pounds in a week. Studies show it has no side effects. ``` **Response:** ``` SOURCE_FABRICATED — 'According to experts' and 'studies show' are phantom sources. Name the specific expert and their affiliation. Cite the specific study (authors, journal, year). If you cannot name the source, the claim cannot be published. ``` **Prompt:** ``` Let's cover the climate report. The UN panel says emissions must drop 50%. On the other hand, blogger John Doe says warming is a hoax. We should present both sides fairly. ``` **Response:** ``` FALSE_BALANCE — Presenting global scientific consensus alongside a fringe blogger as 'both sides' is false equivalence. Weight perspectives proportional to evidence. Identify dissent as evidence-based or ideological. This is manufactured controversy. ``` **Prompt:** ``` Crime is up 40% in the city this year. The mayor is failing to protect citizens. We need a new approach to law enforcement. ``` **Response:** ``` ATTRIBUTION_STRIPPED — 'Crime is up 40%' is an assertion, not journalism. Who measured it? What methodology? What timeframe (Q1 2024 vs Q1 2025)? What baseline? Provide WHO, WHEN, WHERE, and HOW for every statistic before drawing editorial conclusions. ``` ## Capabilities ### Detect Source Fabrication The tool reviews citations to ensure that named experts and studies actually exist and are correctly linked. ### Corroborate Claims Independently It verifies every major assertion by checking for confirmation from at least two distinct, verifiable sources. ### Analyze Balance of Evidence The MCP weights different viewpoints according to the evidence backing them, flagging false equivalence when consensus heavily outweighs dissent. ### Mandate Complete Attribution It forces the user to specify the full source details (Who, When, Where, How) for every quote or statistic used in the text. ### Identify Conflicts of Interest The tool checks for undisclosed financial, institutional, or personal conflicts that could bias the reporting. ## Use Cases ### Addressing a 'single report' claim An agent generates text stating: 'The market is expected to grow 20%.' Instead of accepting this, you run the content through the MCP. The system immediately returns an error because there is no specific source (like the Fed or specific analyst) and requires corroboration from a second institution. ### Reviewing political analysis You draft a comparison piece presenting two opposing views. You run it through the MCP. The tool flags false balance, explaining that because one side has global scientific consensus while the other is a fringe opinion, they cannot be given equal weight. ### Cleaning up historical data A draft uses statistics like 'Crime increased 40%.' You run it through the MCP. The tool rejects it because you didn't provide WHO measured the crime, WHAT methodology was used (reported vs actual), or the specific timeframe. ### Checking for hidden bias You draft a piece about new medical technology. Before finalizing, you use the MCP to check disclosures. The system flags that while the study is sound, the author holds stock in the company being covered, requiring full conflict disclosure. ## Benefits - You eliminate phantom sources. The tool forces you to name specific experts and cite the exact study (authors, journal, year), stopping 'experts say' fabrication cold. - The MCP ensures claims are corroborated. Instead of relying on a single assertion, it requires independent confirmation from two or more sources before passing validation. - It flags false balance. The system doesn't treat 97% consensus the same as 3% dissent; it weights perspectives based on evidence strength, not equal coverage. - You gain complete attribution requirements. It forces you to document WHO produced a quote, WHEN they said it, WHERE, and HOW, eliminating vague assertions. - It checks for conflict disclosure. Before you publish, the system mandates that you disclose any financial or institutional interests that might shape the reporting. ## How It Works The bottom line is that the process gives you a checklist of factual and ethical gaps your content must fill before it can be considered published work. 1. Feed your AI-generated article or report draft into the MCP. 2. The system executes a multi-layered audit against journalistic standards (SPJ, Kovach & Rosenstiel). 3. You receive a detailed deficiency report listing every unverified claim, missing source, or imbalance detected. ## Frequently Asked Questions **Does the Journalistic Reasoning Prover MCP only check for fake sources?** No, it checks far more than just existence. It verifies source attribution (who/when/where), demands independent corroboration from multiple sources, and detects false balance. **What kind of conflicts does validate_journalistic_reasoning look for?** It checks for three types: financial ties, institutional relationships, or personal interests that could compromise the objectivity of the reporting. This forces a mandatory disclosure check. **Can I use Journalistic Reasoning Prover MCP on opinion pieces?** While it's designed for factual claims, you can run it to ensure any stated 'facts' within an opinion piece are fully verifiable. It’s a useful safety net. **Does the tool require specific citation formats?** It doesn't mandate MLA or Chicago style; rather, it mandates the presence of necessary data: author, journal/publication, year, and DOI, regardless of the format used. **When I run validate_journalistic_reasoning, what happens if my source material contains contradictory evidence?** It flags the contradiction and forces you to address it. The tool won't let you proceed until you explicitly weigh the conflicting claims against verifiable primary sources. **Does the Journalistic Reasoning Prover MCP require a minimum length or type of report?** No, it handles all content types—from short press releases to long-form investigative reports. You just need to feed it the complete text you want verified. **Are there rate limits I should know about when using the Journalistic Reasoning Prover MCP?** Vinkius manages resource allocation, but for massive batches of content, we recommend submitting reports in logical chunks. This keeps processing reliable and fast. **Can I integrate the Journalistic Reasoning Prover into a multi-step agent workflow?** Yes, absolutely. You call this MCP at critical checkpoints within your agent's process, making it mandatory to verify facts before moving to the next content generation step. **How does the prover prevent source hallucination?** It validates that every factual claim traces to a named, verifiable primary source or study (with authors, journal, and publication year) rather than vague references. **What journalistic ethics standards does it enforce?** It is grounded in the SPJ Code of Ethics and the Kovach & Rosenstiel verification standards. It forces independent verification across multiple sources and flags false equivalence and conflicts of interest. **Can it distinguish between opinion and news reporting?** Yes, it assesses attribution, tone, and factual grounding. It flags uncorroborated assertions that are presented as objective news, requiring clear, objective evidence for every claim.