neuroterminal

Full financial profile of a US-listed company from official SEC filings in one call: revenue, margins, EPS, assets, debt, cash flow as 5-year series plus computed ratios. The raw SEC source is 3.6MB; this is 7KB. Annual 10-K US-GAAP filers only - foreign private issuers filing 20-F return 404 rather than an empty answer.

Run your own CSS selectors against a JavaScript-rendered page and get back exactly the values you asked for. Precise scraping without pulling a whole page through your context window.

Check a URL before spending tokens fetching it: bot-protection (cloudflare/datadome/perimeterx), JavaScript required, paywall, content length, and robots.txt policy for AI crawlers.

What actually changed in a company's financials since a given date: every metric compared across annual filings, ranked by size of change, with the biggest mover highlighted.

Every recent SEC filing for a US company with direct document links - 10-K, 10-Q, 8-K and the rest - with filing dates, industry code and exchanges. Filter by form type.

Every link on a page, split into internal and external and labelled by zone: nav, footer, or main content. Content links are the ones worth crawling. Optionally JavaScript-rendered.

What prediction markets price for an event, from Polymarket and Kalshi in one call: implied probability, volume, liquidity, close date, plus a cross-venue comparison when the questions genuinely match.

Scan of live prediction markets: which liquid markets sit closest to a coin flip (35-65%) - the genuinely undecided ones - plus any event where Polymarket and Kalshi price the same question differently. Cross-venue gaps need a genuine question match and are often empty; the uncertain-markets list always returns. Filter by minimum volume.

Black-Scholes option price and full Greeks (delta, gamma, vega, theta, rho) with dividend yield, or solve for implied volatility from a price. Exact closed-form maths verified against the standard reference.

Clean readable main text of any page - JavaScript rendered, boilerplate stripped. Cuts your token bill: a 700KB page becomes a few KB of actual content. Works on SPAs plain fetching cannot read.

Risk statistics from a return series: Sharpe, Sortino, Calmar, annualised return and volatility, max drawdown, VaR, CVaR, win rate. Exact arithmetic on up to 10,000 points.

Kelly criterion position sizing: optimal and fractional Kelly, expected value, edge, payoff ratio and break-even win rate. Says plainly when there is no edge.

Normalized product facts from any product URL - name, brand, price, currency, availability, image. Works even when the page has no schema.org markup, using a headless browser fallback.

Fully rendered HTML after JavaScript executes, from a real headless Chrome. For React/Vue/Next SPAs and lazy-loaded content where a plain fetch returns an empty shell. Returns ok, url, html, length and truncated; html is capped at 900,000 chars. Public hosts only.

Schema.org structured data as it exists AFTER JavaScript runs. Many stores inject Product markup client-side, so a plain fetch finds nothing - measured on a third of live Shopify stores.

E-commerce domain intelligence: platform detection, schema.org product markup health, whether the store needs JavaScript rendering, policy pages present, product count estimate.

Which technologies a site actually runs, watched in a real browser instead of guessed from HTML. Static tools miss around 9 runtime-loaded services per site and over-report by 2.5x. Every technology is labelled as observed on the network or merely matched in markup, with the accuracy gap returned as numbers.