Every agent-payable API.

Fetch a PDF by URL and return its text as Markdown-ish plain text, page by page

Fetch a URL with a real rendered browser (JavaScript executed) and return a PDF of the full page. Not JSON -- the response body is the file itself. Results are cached for 24 hours per URL. Cache status and fetch time are reported in the X-Cached and X-Fetched-At response headers.

Parses multi-page PDF documents from URLs, returning structured text content and page counts.

Extract plain text and metadata (title, author, producer, page count) from a PDF supplied by URL or base64 — heavy pdf.js parsing an agent can't self-run.

Extract all text from a PDF. Send as pdf_base64 (base64-encoded PDF, max ~10 MB decoded). Returns text (full concatenated text), pages array (per-page text + char_count), page_count, and metadata (title, author, creator). Encode with: Buffer.from(pdfBytes).toString('base64'). Ideal for RAG pipelines, document QA, or LLM ingestion.

Call when an agent needs text extracted from a public PDF URL (SSRF-safe) or pdf_base64 (8MB cap) as markdown. Exact $0.01 USDC. Prefer unpaid POST /v1/sandbox/pdf-to-md first.

Metadata and structure of a PDF (URL): pages, page size, title, author, creator, dates, version, encryption, form fields, attachments, bookmarks (table of contents) and whether it has extractable text or is a scan.

PDF to text: give a PDF URL, get the text of every page with the title, author and page count. For RAG, research and summarizing reports, filings and papers.

Bounded direct observation of a public company website with hashes, injection screening and signed delivery

Provider-backed entity enrichment with explicit match and limitations

Execute any selected PHION service with a budget, persistent idempotency, acceptance checks and a signed completion envelope

Bounded low-cost inference with a signed delivery receipt

Evidence-grounded research synthesis with citations and signed delivery

Source-returning web search with normalized results and signed delivery

PII redaction for agent pipelines: remove personal data from text before it goes to another model, a log, a ticket system or a dataset. Emails, IBANs (checksum verified), card numbers (Luhn verified), phone numbers and IP addresses are found by exact patterns; names of people, street addresses, birth dates and ID numbers by Claude Haiku 5.5.

Redact personal data from text without AI: emails, phone numbers, IBANs, card numbers (Luhn-checked), Spanish DNI/NIE, IPs, crypto addresses and URLs, replaced by labels ([EMAIL_1]) or asterisks. Before sending text to an LLM or sharing logs. Text is not stored.

PII sanitization — pay-per-call entry point

PR Description From a Diff — Writes a pull request title and description from a pasted unified diff, as Markdown only - a title line of at most 72 characters, then four fixed sections - Summary, Changes, Tests and Risks - where every file the diff touches is named once, including renames, deletions, binary files, mode changes, lockfiles and formatting-only edits, and nothing is claimed that the diff does not show. It describes what the hunks change and leaves out a reason, a speed gain, a secur…

Look up an Italian public administration or public-service operator by 11-digit codice fiscale in IndicePA (AgID). Use when an agent must verify a public-sector counterparty, find its official PEC (certified email) addresses, or confirm it exists before e-invoicing. Input: ?cf=80016350821.

Business Frameworks business-frameworks/growth/decide/pace@1.0 — Set the pace of growth. Call when an agent operating a business (launch to ~$50M revenue) must decide on growth and wants the rule, the inputs it needs, the action and the check. Returns 220 tokens of original synthesis with public citations and child handles; free layer at /catalogue.

Package intelligence across npm, PyPI and crates.io in one call: latest version, license, repository, dependencies and download counts

Health check for an npm or PyPI package before your agent installs it: latest version and release date, weekly downloads (npm), licence, repository, and known security vulnerabilities for a given version.

Check dependencies before your coding agent installs them: up to 20 packages per call (npm, PyPI, crates.io, Go, Maven, NuGet, RubyGems), one decision each: ALLOW, CAUTION or BLOCK with reasons. Catches hallucinated package names that do not exist (slopsquatting), known malware (OSV malicious-package feed, including versions already removed from the registry), typosquats one typo away from a popular package, critical and high vulnerabilities in the exact version with a safe upgrade, deprecated…

Calculates volume, girth and chargeable weight from package dimensions. Missing facts are returned as null and named in unavailable_fields where applicable. Derived values disclose their method; source-backed values disclose source, time window and age.

Retrieves current package metadata from npm, PyPI, crates.io or Maven Central. Missing facts are returned as null and named in unavailable_fields where applicable. Derived values disclose their method; source-backed values disclose source, time window and age.

Call before adding or upgrading a dependency: package-registry metadata for one package (PyPI or npm): latest version, summary, license, dependency count; PyPI also returns 30-day downloads.

Call before adding or upgrading a dependency: package-registry metadata for one package (PyPI or npm): latest version, summary, license, dependency count; PyPI also returns 30-day downloads.

Check npm package metadata and security before installation.

Returns the full version history of a package from a single normalised schema that covers seven ecosystems. Versions are ordered by semantic version, not alphabetically and not in registry order, so 1.10.0 sorts after 1.9.0 — the mistake that quietly breaks naive comparisons. For each version you get the publication timestamp, whether it is the registry default, whether it is deprecated and the deprecation reason as the registry states it.

Package Vulnerability Check (OSV). Package vulnerability check before you install or ship a dependency: known vulnerabilities for a specific version from OSV.dev (GitHub advisories, PyPA, RustSec, Go, npm...) with CVE aliases, severity, fixed-in versions and the recommended upgrade. npm, PyPI, crates, Go, Maven, NuGet, RubyGems, Packagist, Pub, Hex.

Package health: version, age, deprecation

Pad hex to a fixed byte width — Genesis402 / UnyKorn Operator Network

Call when an agent needs a structured SEO/tech audit of a public URL: HTTPS, status, title, meta, robots, canonical, JSON-LD, H1 count, load hints, score 0–100, issues[]. SSRF-safe. Exact $0.01 USDC. Prefer unpaid POST /v1/sandbox/page-audit first.

Page-change monitoring for 30 days: we poll a URL on schedule, you poll us for changes

Machine-readable description audit fetches current public evidence from the supplied target and evaluates the documented criterion. It returns the observation, threshold, verdict, exact evidence source, limitations and a mode-specific action. The check is passive: no login attempt, payload injection, exploit, purchase, form submission or load test is performed.

Machine-readable description decision fetches current public evidence from the supplied target and evaluates the documented criterion. It returns the observation, threshold, verdict, exact evidence source, limitations and a mode-specific action. The check is passive: no login attempt, payload injection, exploit, purchase, form submission or load test is performed.

Machine-readable description repair plan fetches current public evidence from the supplied target and evaluates the documented criterion. It returns the observation, threshold, verdict, exact evidence source, limitations and a mode-specific action. The check is passive: no login attempt, payload injection, exploit, purchase, form submission or load test is performed.

Fetches a URL and returns clean article text/markdown plus metadata. Deterministic, static HTML only, no JS rendering.

Given a set of specific page URLs, find the Google keywords those pages jointly rank for, with each page's position, URL and estimated traffic (ETV) per keyword, plus search volume, CPC, competition, difficulty and intent. Page-level, not domain-level. Optional exclude_pages narrows to keywords the pages rank for that others do not.

Fetch a reviewed public page and return its links with resolved absolute URLs and anchor text, the canonical URL, the declared meta robots directive, and how many links leave the host. Relative hrefs are resolved against the page, so a caller never has to guess a base.

Extracts OG tags, canonical URL, title, and meta description from any web page. Useful for link previews, SEO audits, and content enrichment.

Fetch a reviewed public page and return its title, every meta and Open Graph tag, and the canonical URL. No JavaScript is executed and no script content is read. Returns the licence and creator for the content alongside it, so reuse terms travel with the result.

Page metadata: Open Graph, Twitter card, canonical, feeds, JSON-LD types, language — Genesis402 / UnyKorn Operator Network

Read an allowed public HTTPS page and return a concise English or Japanese report with source quotes, URL and retrieval time. No login-only pages or JavaScript rendering.

Page speed API - Google PageSpeed Insights (Lighthouse) performance audit for a URL on mobile or desktop. Returns the performance score plus lab Core Web Vitals (LCP, FCP, CLS, TBT, Speed Index, TTI) and, where available, real-user (CrUX) field data. Data source: Google PageSpeed Insights.

Fetch a reviewed public page and return its HTML tables as arrays of rows and cells. Bounded to ten tables, fifty rows each and twenty cells per row, and every bound that is reached is named in the warnings so a truncated table is never mistaken for a short one.

Machine-readable title audit fetches current public evidence from the supplied target and evaluates the documented criterion. It returns the observation, threshold, verdict, exact evidence source, limitations and a mode-specific action. The check is passive: no login attempt, payload injection, exploit, purchase, form submission or load test is performed.

Machine-readable title decision fetches current public evidence from the supplied target and evaluates the documented criterion. It returns the observation, threshold, verdict, exact evidence source, limitations and a mode-specific action. The check is passive: no login attempt, payload injection, exploit, purchase, form submission or load test is performed.

Machine-readable title repair plan fetches current public evidence from the supplied target and evaluates the documented criterion. It returns the observation, threshold, verdict, exact evidence source, limitations and a mode-specific action. The check is passive: no login attempt, payload injection, exploit, purchase, form submission or load test is performed.

Website change monitoring for agents that check in a loop: send a URL and get only what changed since your last call (added and removed lines of the page text), plus optional keyword alerts. First call saves a baseline and returns a cursor; pass the cursor on the next call. The cursor only advances when a change is reported, so nothing slips through between polls. Good for price pages, docs and changelogs, job boards, status pages, competitor sites, government notices.