information-retrieval

Fetch a public web page and return clean readable text with its title and metadata - the form a model can actually use, with navigation, scripts and styling removed. Follows up to 3 redirects, caps the download at 2000000 bytes and gives up after 15000ms. Reaches the PUBLIC internet only: the hostname is resolved and every redirect hop re-checked against private, loopback, link-local and CGNAT ranges, so this cannot read internal services or cloud metadata. Returns raw HTML too on request.

Full biomedical research digest — scored papers, MeSH term landscape, and top authors for any medical, clinical, or scientific query from PubMed and OpenAlex

Given a package name and ecosystem, query OSV for known vulnerabilities and get a pinned-or-upgrade verdict an agent can act on before adopting a dependency.

Computes a SHA-256 digest of any input text and returns it as a hex string, computed locally with no network round-trip to a third party. Also supports SHA-512, SHA-1, and MD5 on request. Useful for content fingerprints, cache keys, and deduplication logic where the same input must always produce the exact same output.

Returns the latest version, summary, license, and required Python version constraint for any package on the Python Package Index, plus its declared dependencies and yank status. A quick pre-install check for coding agents working in Python before they commit to a dependency.

Parses loosely formatted phone number input and returns its E.164 standard representation, country, and line type, with a default-country hint for numbers missing a country code. Built for lead deduplication, SMS delivery systems, and validation steps that reject malformed numbers before they hit a paid messaging API.

Country reference data keyed by name or ISO code — currency, capital, languages, calling code, region, area, and neighboring countries — returned as one clean JSON object. Useful for populating currency selectors, tax jurisdiction logic, and data-enrichment pipelines without hardcoding a static country table.

Returns normalized metadata for any npm package, current version, license, dependency list, and whether it's been deprecated, without needing to parse the full, much larger raw registry response. For agents auditing a project's dependency tree or deciding what to install.

Reads a public PDF and returns its text content with structure preserved, handling standard digital PDFs up to 25MB. Files are processed in an isolated sandbox and discarded immediately. Useful anywhere a workflow needs a PDF's contents as plain text instead of a downloadable binary.

Generates or validates universally unique identifiers (UUIDs).

Returns rhyming words, sound-alikes, or commonly paired adjectives for any input word, ranked by strength of association. Useful for crossword and word-game agents, poetry generation, and naming brainstorms that need a wider candidate list than a model would produce unprompted.

Finds words related to any input word, similar meanings, words that sound alike, or adjectives commonly paired with it, ranked by association strength, up to 20 results per call. For copywriters, product namers, and word-game agents looking for options beyond an LLM's usual first guesses.

Performs DNS lookups to resolve domain names to their IP addresses.

Retrieve an attestation by id with GET /attest/{id}. The gateway prefix-matches this registered path.

Social share metadata lookup: reads a web page's title, canonical link, favicon, author, and social-card tags (Open Graph and Twitter formats) and returns them as one normalized JSON object. Safe against internal-address requests. Handy for SEO checks, content management systems, and any tool that needs to know how a URL will look when shared.

Read a URL as clean markdown/text for LLMs. Append the full target URL to the path, e.g. /https://example.com. Returns the extracted title, source URL and markdown content as plain text.

RFC 5322 syntax check, live MX resolution and a disposable-provider check for one email address, in a single call.

Performs web searches and returns results from indexed online sources.

Returns the activity rollup for one jurisdiction, including total WARN events on file — use it to check disruption volume before drilling into regional or county detail

Fetch a public web page and return it as clean Markdown with its title. Capped at 2 MB.

Verifies an ICD-10-CM diagnosis code or searches the code set by text, returning official code descriptions.

Searches PubMed and returns citation metadata for each hit: title, journal, publication date, authors, DOI and PMID.

Search the National Vulnerability Database by keyword, CPE, publication window or CVSS severity, returning a paginated list with scores and CWE classifications.

Screens open ClinicalTrials.gov studies for a condition against a patient age, sex and healthy-volunteer status, returning the trials the structured criteria do not exclude with their inclusion and exclusion criteria split out.