Building an MCP Server That Verifies Its Sources: Inside footnote-mcp

A developer built footnote-mcp, a Python MCP server that verifies source claims using heuristic and LLM-based backends. The tool achieves 100% accuracy on numeric and factual data claims and 83% overall accuracy, with a blind spot in semantic negation and paraphrase. It supports web search, page fetching, and browser automation without requiring API keys.

footnote-mcp is a Python MCP server installable via pip, Docker, or pipx. No API keys required — it falls back to scraped Bing + DuckDuckGo search and automatic headless Chromium for JavaScript-heavy pages. The core tool is evidence entailment . It takes a claim and a source text, and returns whether the claim is supported, unsupported, or contradicted. The heuristic backend extracts numeric and named-entity tokens from both the claim and source, then checks for exact matches and contradictions. On its design domain — numeric and factual data claims — it achieves 100% accuracy on a labeled benchmark set. For semantic cases negation, paraphrase , the ollama backend uses a local LLM as a judge. Three tools build on this: corroborate claim triangulates a claim across multiple sources, locate claim span finds the exact supporting sentence with character offsets, and build research debug report produces a compact report of queries, URLs, and verification gaps. web read fetches pages through a 5-tier escalation ladder: HTTP curl cffi to rotating proxy to headless Chromium to Chromium through proxy to hosted scrape API Firecrawl/ScrapingBee . A block/quality detector decides when to escalate, and per-domain rate limiting, circuit breakers, and negative cache keep it polite. web search supports Tavily, Brave, Google, or scraped Bing + DuckDuckGo as fallback. Pass semantic: true to reorder results by meaning using local Ollama embeddings. Beyond text, the server handles tables, CSV/XLSX/PDF/JSON, date validation, unit resolution, and time series reconciliation. For JavaScript-heavy pages, 10 browser tools let you drive a headless Chromium session. When generic parsers fail, the server can synthesize sandboxed extraction code through a controlled recipe system. The heuristic backend achieves 100% accuracy on numeric and factual data claims n=15 . Overall accuracy including semantic cases is 83% n=18 . The blind spot is purely semantic negation and paraphrase — the ollama backend closes that gap. pip install footnote-mcp python -m playwright install chromium footnote-mcp Then add to your MCP client config and start researching. Full source at github.com/KazKozDev/footnote-mcp https://github.com/KazKozDev/footnote-mcp . Feedback welcome — especially on the verification approach. The benchmark is designed to be extended, so if you have a claim/source pair that should be caught, open an issue or PR.