Both tools return source distributions, political-leaning breakdowns and sample headlines next to the score, so you can see whether a reading rests on ten articles or a thousand. It reads US newsrooms only — treat it as a US-market signal rather than a global one.
A Node server over EventRegistry. It samples curated US newsrooms, filters for investor relevance, and returns two sentiment scores on a 0-10 scale with the sampling diagnostics behind them.
- A single day's sentiment snapshot, taking a plain date or natural language such as yesterday or last Friday — analyze_headlines
- Months aggregated between a start month and an end month — analyze_monthly_headlines
- Two scores per result, general and investor, normalised to a 0-10 scale
- Political-leaning breakdowns, source distributions and sample headlines alongside the score
- Structured JSON compatible with MCP structuredContent, plus token budgeting and rate-limit telemetry
Node.js v18 or later and an EventRegistry API key in NEWS_API_KEY. Build with npm install and npm run build, then run build/index.mjs with TRANSPORT=stdio. For HTTP, set TRANSPORT=http with HOST and PORT, and optionally ALLOWED_HOSTS and ALLOWED_ORIGINS; a GET on /healthz returns 200 ok.
Build from source — clone the repository and build it, then point your client at the binary