openverbs.com

Validate a 9-digit US ABA bank routing number by its weighted (3,7,1) mod-10 checksum. Spaces and hyphens are ignored.

Add an ISO 8601 duration to a start date or date-time and return the resulting UTC instant. Calendar-aware: months and years honor month lengths (e.g. Jan 31 + P1M clamps to the last day of February). A negative-signed duration subtracts.

Normalise an address to its canonical form — Ethereum to its EIP-55 checksummed form, bech32 Bitcoin to lowercase (base58check is left untouched as it is case-sensitive). An address that is not a valid address of a known chain is rejected before payment.

Validate a blockchain address. Supply `chain` to check against a specific chain, or omit it to auto-detect. Ethereum addresses report their EIP-55 checksum state; Bitcoin covers base58check (P2PKH/P2SH) and bech32/bech32m segwit (P2WPKH/P2WSH/P2TR). A well-formed but invalid address returns `valid:false` — not an error.

Adler-32 checksum (as used by zlib) over the input payload. Faster than CRC-32 but weaker for short inputs. Returns the checksum as an unsigned integer and 8-digit hex. Non-cryptographic.

Split an amount into parts using integer minor-unit math so the shares always sum back to the exact original — no rounding drift. Pass parts for an even split, or ratios for a weighted one (e.g. [1,1,1] splits $100 into 33.34/33.33/33.33). The leftover minor units go to the largest fractional remainders.

Score how likely a URL is to be phishing or untrustworthy using offline lexical and structural heuristics (no network lookups). Checks raw-IP hosts, embedded credentials, punycode/homoglyph labels, brand typosquatting, brand-in-subdomain, abused TLDs, shorteners, random hostnames, odd ports, plaintext http and executable paths. Returns a 0-100 risk score, a verdict and the reasons.

Inspect a non-negative integer's bit structure: bit length, set-bit count (population count), trailing-zero count, power-of-two test, and hex/binary representations. An optional width zero-pads the binary form.

Compute the astronomical Julian Date (JD, fractional, epoch noon 1 Jan 4713 BC) and Modified Julian Date (MJD = JD − 2400000.5) of a UTC instant. Unlike `calendar`'s integer Julian Day Number, this carries the time-of-day fraction.

Determine the astronomical season for a date, based on the Sun's apparent ecliptic longitude (0°=March equinox, 90°=June solstice, …). Defaults to the northern hemisphere; pass `hemisphere` to flip.

Balance an equation such as "H2 + O2 -> H2O" into the smallest whole-number coefficients using exact rational linear algebra. Equations that cannot be balanced with positive coefficients are clean 400s.

Binomial distribution with `trials` (n ≥ 0, ≤ 1e6) and success `probability` (0..1). `function` selects the mass (pmf), cumulative distribution (cdf) or quantile / inverse-CDF.

Circularly rotate a non-negative integer left or right within an explicit bit width (bits shifted off one end reappear at the other). The value must fit in `width` bits. Returns decimal, hex and binary.

Shift a non-negative integer left or right by `amount` bits. A right shift is logical (fills with zeros). A left shift grows freely unless a `width` is given, in which case the result is masked to that width and the input must fit it. Returns decimal, hex and binary.

Bitwise AND across two or more non-negative integers (given as decimal strings, evaluated as arbitrary-precision BigInt). Returns the result in decimal, hex and binary.

One's-complement NOT of a non-negative integer within an explicit bit width (two's-complement domain). The value must fit in `width` bits. Returns decimal, hex and binary.

Bitwise OR across two or more non-negative integers (given as decimal strings, evaluated as arbitrary-precision BigInt). Returns the result in decimal, hex and binary.

Bitwise XOR across two or more non-negative integers (given as decimal strings, evaluated as arbitrary-precision BigInt). Returns the result in decimal, hex and binary.

Break a braille string into its cells and the raised-dot numbers (1–8) of each cell.

Transcribe a Unicode braille string back to text, honouring capital and number signs. Non-braille or unmapped cells are clean 400s.

Build a chord from a root note and quality (major, minor, dim, aug, maj7, min7, dom7, dim7, half-dim7, sus2, sus4, maj6, min6). Notes are spelled by stacked thirds (C dom7 → C E G Bb). Returns the note names, MIDI numbers and semitone intervals.

Build a scale from a root note and type (major, natural-minor, harmonic-minor, melodic-minor, dorian, phrygian, lydian, mixolydian, locrian, major-pentatonic, minor-pentatonic, blues, chromatic). Heptatonic scales and modes are spelled diatonically (F major → F G A Bb C D E). Returns the note names, MIDI numbers and semitone intervals.

Assemble an ISO 8601 duration string from individual components (years, months, weeks, days, hours, minutes, seconds) and an optional sign. All components must be non-negative and at least one must be non-zero.

Add/subtract business days to a date, or count business days between two dates — skipping weekends and optional holidays.

Test whether an IP address falls within a CIDR block. A well-formed address of a different family (or simply outside the block) returns contains=false — not an error.

CRC-16 over the input payload in one of several standard variants (ccitt-false, xmodem, modbus, ibm; default ccitt-false). Returns the checksum as an unsigned integer and 4-digit hex. Non-cryptographic.

CRC-32 (IEEE 802.3 polynomial, as used by zip, gzip and PNG) over the input payload. Returns the checksum as an unsigned integer and 8-digit hex. Non-cryptographic — for error detection only.

Parse a document (in any format) and re-emit it in the RFC 8785 JSON Canonicalization Scheme form — sorted keys, canonical number/string formatting — plus the SHA-256 of the canonical bytes for stable hashing, signing and dedup.

Split an identifier or phrase into words and render it in every common case — camelCase, PascalCase, snake_case, kebab-case, CONSTANT_CASE, dot.case, path/case, Title Case and Sentence case. Handles acronyms (HTMLParser → html parser).

The n-th Catalan number Cₙ = C(2n, n)/(n+1), counting balanced bracketings, binary trees, and many other structures. Requires n ≤ 10000.

Guess which chain an address belongs to from its shape and encoding, without requiring the caller to know it up front. Returns the chain, format and address type, or nulls when nothing matches.

Check whether a specific date is a public holiday in a country (optionally by state/region).

Given a reaction and the supplied amount of every reactant, identify the limiting reagent and compute leftover reactant and theoretical product yields.

Generate a harmony scheme (complementary, analogous, triadic, tetradic, monochromatic) from a base color.

Binomial coefficient nCr = n!/(r!(n-r)!), the number of unordered r-subsets of n. Requires r ≤ n and min(r, n-r) ≤ 10000.

Compare two semantic versions and report which is greater (precedence-aware).

Summarise a text's difficulty as a single consensus US grade level (averaging the five grade formulas), plus the Flesch Reading Ease, a plain-language difficulty band and an approximate age range.

Rebase a timecode (or a raw frame count) from one frame rate to another on a real-time basis, so the wall-clock position is preserved. Give a source frameRate plus timecode or totalFrames, and a toFrameRate (with optional toDropFrame). Returns the target timecode, frame count and seconds.

Parse any CSS color and return it as hex, rgb, hsl, hwb, lab, lch and oklch.

Transcode a document between json, json5, jsonc, yaml, xml, csv, tsv, toml, ndjson, ini, env, properties, plist and querystring. Any source converts to any target through a JSON hub. Malformed input or a target that cannot represent the structure is rejected before payment.

Convert an ISO datetime into another timezone, with offset, weekday and Unix time.

Count syllables per word and in total across the text.

Choose the singular or plural form for a given count, e.g. (apple, 3) → "apples", (apple, 1) → "apple". By default the count is prepended ("3 apples"); set includeCount=false to omit it.

Build the flag emoji for an ISO 3166-1 alpha-2 country code from its regional-indicator symbols (e.g. "US" → 🇺🇸). Non two-letter codes are clean 400s.

Return the validation regular expression (source and flags) used for a country's postal codes.

Build a valid .ics document from event fields (summary, start/end, location, …).

Convert text into a URL-safe slug: transliterate accents, apply a locale character map, and join with a separator.

Damerau-Levenshtein distance — like Levenshtein but also counts adjacent transpositions as a single edit — plus a normalized similarity in [0,1].

Return the ISO weekday (1=Monday..7=Sunday) and its name for a Gregorian date.

Return the ordinal day of the year (1..366), the number of days in that year and whether it is a leap year.

Convert a decimal string to an exact reduced fraction. For a repeating decimal, set repeat to the number of trailing fractional digits that repeat (e.g. value='0.1666', repeat=1 → 1/6; value='0.3', repeat=1 → 1/3). A non-numeric value is a clean 400.

Decode international Morse code back into text. Letters are separated by whitespace and words by "/". Unknown Morse sequences are clean 400s.

Read the text out of a QR code or barcode in a PNG or JPEG image. Auto-detects QR, DataMatrix, PDF417, Aztec and 1D symbologies (Code128/39/93, EAN-13/8, UPC-A/E, ITF, Codabar). A valid image with no code returns { found: false }; a payload that is not a parseable image is a 400.

Turn a string of phonetic words back into text ("Sierra Oscar Sierra" → "SOS"). Accepts common alternative spellings (Alpha, Juliet, Niner). Unknown words are clean 400s.

Find personally identifiable information in a text blob: emails, phone numbers, credit cards (Luhn-validated), IBANs (mod-97 validated), IPv4/IPv6 addresses, US SSNs and JWTs. Returns every match with its type and character offsets, plus per-type counts. Validated — not just pattern-matched — to keep false positives low. Fully offline.

Guess the character set of a base64-encoded byte buffer: sniffs a byte-order mark, then falls back to ASCII / UTF-8 validation and a UTF-16 NUL-pattern heuristic. Returns a best-effort charset with a confidence label — a low-confidence guess is still a successful 200.

Detect the issuing network/brand (Visa, Mastercard, American Express, Discover, Diners Club, JCB, UnionPay, Maestro) from the card's IIN prefix, with the network's valid lengths. Returns `network:null` for an unrecognized prefix.

Sniff which structured format an unknown document is. Because formats overlap, returns every plausible format with a 0–1 score plus the best guess — honest about ambiguity rather than pretending certainty.

Compute the determinant of a square matrix via Gaussian elimination with partial pivoting. A non-square, ragged or empty matrix is a clean 400; a singular matrix simply returns 0.

Sørensen-Dice similarity coefficient in [0,1] over character bigrams — robust to word order and minor edits.

Look up a domain's RDAP registration data — registrar, status, creation/expiry/update dates, nameservers and DNSSEC. Unregistered domains return found:false.

Build braille cells from raised-dot numbers (each cell an array of 1–8). Out-of-range or repeated dots are clean 400s.

Encode a word with Double Metaphone, returning a primary and a secondary code to capture alternate pronunciations (useful for non-English names).

Compute the calendar breakdown (years, months, days, hours, minutes, seconds) between two ISO 8601 dates/date-times, plus exact totalSeconds. If end precedes start the result is flagged negative rather than erroring.

Validate a GTIN barcode number — EAN-8, UPC-A (12), EAN-13 or GTIN-14 — by the standard mod-10 check digit. Length selects the scheme.

Apply the EIP-55 mixed-case checksum to an Ethereum address, regardless of its input casing. Returns the checksummed address and whether it differed from the input. Non-hex input is rejected before payment.

Look up a periodic-table element by exactly one of `symbol` (e.g. "Fe"), `name` (e.g. "iron") or atomic `number` (1..118). Returns its number, symbol, name, standard atomic weight and whether that weight is a true standard atomic weight.

Search the periodic table by name or symbol substring. Exact matches rank first, then prefix matches, then substring matches. Returns up to `limit` elements (default 20) plus the total match count.

Encode text into international (ITU) Morse code. Letters are separated by spaces and words by "/". Supports A-Z, 0-9 and common punctuation; unsupported characters are clean 400s.

Encode a coordinate to a geohash and return its bounding box and 8 neighbors.

Evaluate a mathematical expression with optional named variables — arithmetic, comparison and logic operators plus functions like sqrt, abs, min, max, sin, cos, log, floor, round, etc. A syntax error, an undefined variable, or a non-finite result is a clean 400. The factorial operator is disabled for safety.

Compute the exact probability distribution of a dice notation — mean, variance, standard deviation, min/max and the full probability mass function — with no sampling. Notations whose outcome range is too large for exact computation are clean 400s.

Return both the fully-expanded (eight 4-hex groups) and canonical compressed (RFC 5952) forms of an address. IPv4 is returned unchanged in both fields.

Expand an RRULE recurrence rule into its occurrence datetimes, with a human-readable description.

Exact factorial n! for 0 ≤ n ≤ 20000, returned as a decimal string with its digit count.

Read a single XBRL financial concept for a company (by CIK or ticker) from its as-filed statements. Give the concept's XBRL tag (e.g. Revenues, NetIncomeLoss, Assets, EarningsPerShareDiluted) and optionally the taxonomy (default us-gaap, also dei). Returns the label and, per unit (USD, shares), recent datapoints with period end, value, fiscal year/period, form, filing date and accession. Bounded, as-reported output.

List a company's most recent EDGAR filings, identified by CIK or ticker. Optionally filter to a single form type (e.g. 10-K, 10-Q, 8-K, 4, S-1). Each filing includes its accession number, form type, filing and report dates, the primary document name, whether it carries XBRL, and a direct URL to the as-filed document. Also returns company name, tickers, exchanges and SIC industry code.

One-call financial snapshot for a US public company (by CIK or ticker), from its as-filed SEC XBRL statements: revenue, cost of revenue, gross profit, operating income, net income, EPS (basic/diluted), total assets, total liabilities, stockholders equity, cash, operating cash flow and shares outstanding. Each line item gives the latest full fiscal year and the latest reported value. Unknown company returns found=false.

Find the first match (or all matches with all=true) of a RE2 pattern, returning each match's text, start/end offsets, and numbered and named capture groups. No match is a successful 200 with an empty match list.

Fletcher checksum over the input payload, in 16-bit (default) or 32-bit width. Returns the checksum as an unsigned integer and hex. Non-cryptographic.

Format a numeric amount as currency for a given locale using Intl, honouring the currency's own minor-unit convention (USD→2, JPY→0). display selects the symbol, narrow symbol, ISO code or full name. An unknown currency or locale is a clean 400.

Project WGS84 longitude/latitude (degrees) to Web Mercator (EPSG:3857) easting/northing in metres. Latitude is clamped to ±85.0511° (the projection's valid extent).

Add, subtract, multiply or divide two fractions and return the exact, fully-reduced result. All computation is done in BigInt so there is no floating-point error. Dividing by a zero fraction is a clean 400.

Expand a fraction to its decimal form by exact long division, detecting a repeating cycle and rendering it in parenthesis notation (e.g. 1/6 → '0.1(6)', 1/7 → '0.(142857)'). maxDigits caps a non-repeating expansion. A zero denominator is a clean 400.

Convert a Julian Day Number back to a proleptic-Gregorian date, with the ISO weekday (1=Monday..7=Sunday) and its name.

Convert an MGRS grid reference back to decimal lat/lon on WGS84 (the coordinate of the grid cell's south-west corner at the given precision). Whitespace is ignored; a malformed reference is a clean 400.

Convert a UTM coordinate (zone, hemisphere, easting, northing) back to decimal lat/lon on WGS84.

Future value of a present sum and/or a level annuity, given a per-period rate and number of periods. Set dueAtBeginning=true for an annuity-due (payments at the start of each period).

Render text (URL, vCard, Wi-Fi config, …) as a QR code — SVG string or PNG data URL.

Generate a random Luhn-valid TEST card number for the given network (default Visa). For testing only — these are not issued cards and cannot be charged. The response is always flagged `test:true`.

Generate a random EUI-48 MAC address. By default it is locally-administered and unicast (safe for testing); set `local:false` for a universal-bit address or `multicast:true` for a group address.

Render a barcode (Code128, Code39, EAN-13/8, UPC-A/E, ITF-14, DataMatrix, PDF417) as SVG or PNG data URL.

Image search API - live Google Images results for a query, localised by location and language. Returns ranked images with title, direct image URL, source page URL and alt text. A google images / image search api for agents.

Distance and initial bearing between two coordinates.

Compute the date of Western (Gregorian) Easter Sunday for a year in 1583..9999 using the Anonymous Gregorian computus.

Report the ground resolution (metres per pixel, for 256-px tiles) and the OGC map scale denominator at a zoom level, optionally adjusted for latitude (default equator).

Format a card number into the conventional display groups for its network (4-6-5 for Amex, 4-6-4 for 14-digit Diners, otherwise groups of four).

Compute a keyed HMAC of some data with a secret key — the standard way to sign and verify webhooks (Stripe, GitHub, etc.) and API requests. Data and key may each be utf8/hex/base64/base64url; the tag is returned as hex/base64/base64url.

Decode HTML entities (named, decimal and hex numeric references) back into plain text.

Encode text to HTML entities so it can be safely embedded in markup. By default uses hex numeric references; set useNamedReferences for named entities (&, ©) or encodeEverything to encode every character.

Hamming distance — the number of positions at which two EQUAL-LENGTH strings differ — plus a normalized similarity. Strings of unequal length are a clean 400.

Compute a cryptographic digest (SHA-2, SHA-3, BLAKE2, and legacy SHA-1/MD5/RIPEMD-160) of some data. The input may be utf8/hex/base64/base64url and the digest is returned as hex/base64/base64url.

Render an ISO 8601 duration as an English phrase, e.g. 'P1Y2M' → '1 year and 2 months'. Components are shown in descending order with correct pluralization.

Validate a 15-digit mobile device IMEI by its Luhn check digit. Spaces and hyphens are ignored.

Validate an ISBN-10 (mod-11, trailing X allowed) or ISBN-13 (GTIN mod-10) book number. Length after removing spaces/hyphens selects the scheme.

Look up a country by its ISO 3166-1 alpha-2 code, alpha-3 code, numeric code or English name. Provide exactly one field. Returns all codes plus the UN region.

Look up an active ISO 4217 currency by its alphabetic code or numeric code. Provide exactly one field. Returns the code, numeric code, name and number of minor units (decimal places).

Look up a language by its ISO 639-1 alpha-2 code, ISO 639-2 alpha-3 code (bibliographic or terminologic) or English name. Provide exactly one field.

Return the ISO-8601 week-numbering year, week number (1..53) and weekday (1=Monday..7=Sunday) for a Gregorian date. The ISO year can differ from the calendar year around New Year.

Infer a JSON Schema (draft-07) from a sample document in any supported format. Types are inferred, present keys are marked required, and heterogeneous array item schemas are merged. The result validates out of the box with /v1/validate.

Infer a JSON Schema (draft-07) from an example value, walking objects and arrays recursively. Set detectFormat=true to tag strings that look like date-time/date/email/uri/uuid, requireAll=false to leave object properties optional, and additionalProperties=true to allow extra keys.

Resolve a single code point (given as 'char' or 'codepoint') to its name, general category, block, script and plane. Provide exactly one of the two inputs.

Report the properties of a MAC address encoded in its first octet: unicast vs multicast (I/G bit), universal vs locally-administered (U/L bit), the broadcast address, and — for EUI-48 — the OUI and NIC portions.

Given an emoji character (or any string), return its code points and — if known — its name, category, shortcodes and tags. An unknown emoji is a successful result with `known:false`, not an error.

Break a string into its Unicode code points, each with its U+ notation, decimal value and UTF-8 / UTF-16 byte layout, plus total byte lengths.

Fetch any URL and, without paying, decode its x402 402 Payment Required challenge: price (atomic and formatted for known stablecoins), asset, network, payTo, resource, description, timeout and any input schema. Flags MPP and Bazaar discovery extensions. Returns free:true when no payment is needed. SSRF-hardened: public hosts only; private, loopback, link-local and cloud-metadata addresses are rejected.

Internal rate of return (per period) of a series of cash flows, solved with a bounded bisection. Requires at least one positive and one negative flow; flows with no sign change or no bracketable root are rejected as a clean 400.

Name the interval between two notes (`to` at or above `from`) — its size in semitones, its diatonic number and its quality (perfect/major/minor/augmented/diminished). Correct spelling matters: C→F# is an augmented fourth, C→Gb a diminished fifth.

Un-project Web Mercator (EPSG:3857) easting/northing in metres back to WGS84 longitude/latitude in degrees.

Invert a square matrix via Gauss-Jordan elimination. A non-square, ragged or empty matrix is a clean 400. A singular (non-invertible) matrix is NOT an error — it returns a successful { invertible:false, inverse:null }.

Jaccard similarity index in [0,1] — the size of the intersection over the union of the two strings' character-bigram sets.

Jaro similarity and the Jaro-Winkler variant (which boosts scores for a shared prefix), both in [0,1]. The prefix boost factor is configurable (default 0.1, max 0.25). Well suited to short strings like names.

Levenshtein edit distance (insertions, deletions, substitutions) between two strings, plus a normalized similarity in [0,1]. Operates on Unicode code points.

Parse an expression and return the free variables it references, without evaluating it — so an agent can discover which inputs a formula needs before supplying values. A syntax error is a clean 400.

List a country's holidays for a calendar year (optionally by state/region, language and type).

List every ISO 3166-1 alpha-2 country code for which a postal-code pattern is available.

Live Google local-finder (map places) results for a query, localised by location and language. Returns ranked places with title, address, website domain, phone, rating and review count.

Search Google's local business directory within a radius of a coordinate, filtered by category and/or a name query. Returns ranked businesses with name, category, star rating and review count, address, phone, website domain, place_id/CID and coordinates. Richer than a SERP local pack. No match → empty results.

Resolve a Unicode block name (loose matching) to its range, size, count of assigned characters and a small sample. An unknown block is a successful `found:false`. This is a Unicode Character Database block — not a blockchain block.

Resolve a Unicode script name (loose matching) to its code point ranges, total size and a small sample. An unknown script is a successful `found:false`.

Resolve a 20-character Legal Entity Identifier (LEI, ISO 17442) to its authoritative entity record: verified legal name, registration status (ISSUED/LAPSED/RETIRED), entity status, jurisdiction, legal form, registration number, legal and headquarters addresses, key dates (registration, last update, next renewal), managing LOU, corroboration level and any BIC codes. ISO 7064 check digits are verified before payment.

Resolve an emoji by its shortcode (e.g. "smile" or ":smile:") or exact name. Provide exactly one field. Returns the emoji, its metadata and code points.

Reverse lookup: resolve a Unicode character name (e.g. 'SNOWMAN', 'CJK UNIFIED IDEOGRAPH-4E2D', 'HANGUL SYLLABLE GA') to its code point. Loose whitespace/case matching. An unknown name is a successful `found:false`.

Resolve a stock ticker symbol or company name to its Central Index Key (CIK) - the identifier every other EDGAR endpoint keys on. Exact ticker matches rank first, then company-name prefix/substring matches. Returns each match's zero-padded CIK, numeric CIK, ticker and official company title.

Resolve a filename, path or extension to its MIME type, charset and full Content-Type.

Validate a number by the Luhn (mod-10) algorithm — the scheme behind credit/debit card numbers and many IDs. Spaces and hyphens are ignored. A mismatched check digit is a successful { valid: false }.

Derive the modified EUI-64 interface identifier from a 48-bit MAC (insert FFFE, flip the U/L bit) and the resulting IPv6 link-local address (fe80::/64).

Resolve a MIME type to its default file extension.

Mask a card number for display, keeping the first six and last four digits (the PCI-permitted maximum) and replacing the rest with a mask character (default `*`).

Encode a word with the Metaphone algorithm — more accurate than Soundex for English pronunciation.

Convert decimal lat/lon to a Military Grid Reference System (MGRS) string on WGS84. Precision is the number of digits per axis (1–5; 5 = 1-metre resolution, default 5).

Compute the molar mass (g/mol) of a chemical formula. Supports nested parentheses and brackets, element multipliers, and hydrate dot notation with coefficients (e.g. "Ca(OH)2", "CuSO4·5H2O", "K4[Fe(CN)6]"). Returns the total mass and the per-element composition with mass fraction percentages.

Compute the Moon's phase name, age in days since the last new moon and illuminated fraction (0..1) for a UTC date/time. Uses the mean synodic month (29.53 days).

Multinomial coefficient (k1+k2+…)! / (k1! k2! …) — the number of ways to partition n = Σki items into labelled groups of the given sizes. The sum must be ≤ 20000.

Multiply two matrices A·B. The inner dimensions must agree (columns of A equal rows of B); a mismatch or a ragged/empty matrix is a clean 400.

Find the nearest CSS named color to a given color, by perceptual distance.

Net present value of a series of cash flows discounted at a per-period rate. cashflows[0] is the time-0 flow (typically the initial outlay, negative).

News search: live Google News results for a query, localised by location and language. Returns ranked articles with title, URL, domain and, where available, source and timestamp. Current events and headlines for agents.

Normal distribution with optional mean (default 0) and standard deviation sd (default 1). `function` selects the density (pdf), cumulative distribution (cdf) or quantile / inverse-CDF.

Normalise a postal code's casing and whitespace (uppercase, collapse runs of spaces) and report whether the normalised form is valid for the country.

Canonicalize a URL: WHATWG parsing already lowercases the host, resolves dot-segments and drops default ports; on top of that, query parameters are sorted by default and you can optionally drop the fragment or a trailing slash. Returns the normalized URL and whether it changed.

Parse a note in scientific pitch notation (e.g. "A4", "C#5", "Bb3") into its letter, accidental, pitch class (0..11), octave, MIDI number and frequency in Hz (A4 = 440 Hz).

Spell an integer out in English words, e.g. 123 → "one hundred twenty-three". Negatives are prefixed with "minus".

Format a non-negative integer as an ordinal — both the suffixed form (21st) and the ordinal words (twenty-first).

Parse an IPv4 or IPv6 address and classify it — version, normalized form and RFC range (unicast, private, loopback, linkLocal, multicast, uniqueLocal, ipv4Mapped, …). Detects IPv4-mapped IPv6.

Parse a MAC / EUI-48 or EUI-64 address from any common notation (colon, hyphen, Cisco dot, or bare hex) and return it in every canonical form plus its integer value.

Parse an absolute URL into its WHATWG components: protocol/scheme, credentials, host/hostname/port, origin, pathname (plus path segments), search (plus a decoded query object) and hash. A string that is not a valid absolute URL is rejected as a clean 400.

Parse a coordinate string into decimal lat/lon. Accepts a plain decimal pair ('40.4462, -79.9822') or DMS with hemisphere letters ('40°26'46"N 79°58'56"W'). Anything unrecognized is a clean 400.

Parse an ISO 8601 duration such as 'P1Y2M10DT2H30M' into its components. Returns a normalized ISO string and totalSeconds — but totalSeconds is null when the duration contains years or months, because those are calendar-variable (use /v1/add or /v1/between for exact date math).

Parse dice notation such as "3d6+2", "4d6kh3" (keep highest 3) or "2d20dl1" (drop lowest 1) into structured terms. Malformed notation is a clean 400.

Split YAML front-matter from a document — returns the parsed metadata object and the remaining content body. Malformed YAML is rejected as a clean 400.

Parse an .ics document into structured calendar events (summary, times, location, RRULE, attendees).

Parse a formatted money string back into a number, using the locale's grouping and decimal separators (so '1.234,56' in de-DE → 1234.56). Currency symbols/codes and whitespace are stripped; parentheses or a leading minus mean negative. A string with no numeric amount is a clean 400.

Parse an SMPTE timecode (HH:MM:SS:FF, or HH:MM:SS;FF for drop-frame) at a given frame rate into its components, total frame count and real-time seconds. Drop-frame is auto-detected from a ';' separator unless dropFrame is set. Out-of-range or non-existent drop-frame values are a clean 400.

Parse a semantic version into major/minor/patch, prerelease and build metadata.

Number of ordered arrangements of r items from n, nPr = n!/(n-r)!. Requires r ≤ n and r ≤ 10000.

Report whether two tokens sound alike under a chosen algorithm (soundex, metaphone or double-metaphone; default double-metaphone). Returns each token's codes and a boolean match.

Look up a business's full Google Business Profile by name and location. Returns name, category, star rating and review count, address, phone, website, opening hours by day, Google place_id and CID, coordinates and claimed status. For place enrichment, store-hours and business lookups. Unknown business returns found=false.

Return the plural form of an English word (handles irregulars — person → people, mouse → mice).

Test whether a coordinate falls inside a polygon.

Poisson distribution with rate `lambda` (> 0, ≤ 1e9). `function` selects the mass (pmf), cumulative distribution (cdf) or quantile / inverse-CDF.

Present value of a future sum and/or a level annuity, given a per-period rate and number of periods. Set dueAtBeginning=true for an annuity-due.

Two directions over query strings. action='parse' decodes a query string (with or without a leading '?') into an object, collapsing repeated keys into arrays. action='build' encodes a params object into a query string, expanding array values into repeated keys; set sort=true to order keys alphabetically.

Convert an integer between numeral bases 2–36 (binary, octal, hex, base36, …). Arbitrary size via BigInt; digits are case-insensitive and a leading minus is allowed.

Test whether a version satisfies a semver range (^, ~, x-ranges, ||, hyphen).

Given a reaction and a known amount (moles or grams) of one species, compute the amount of a target species. The equation is balanced automatically.

Compute the standard English readability formulas — Flesch Reading Ease, Flesch–Kincaid grade, Gunning fog, SMOG, Coleman–Liau and ARI — along with the underlying sentence/word/syllable statistics. Empty text is a clean 400.

Estimate reading and speaking time for a text at configurable words-per-minute rates (defaults: 200 wpm reading, 130 wpm speaking).

Return the input text with every detected PII span replaced. `mode` controls the replacement: "label" (default) → `[EMAIL]`; "mask" → the value covered with a mask character; "hash" → `[EMAIL:1a2b3c4d]`, a stable per-value fingerprint so equal values stay linkable without exposing them. Same validated detectors as `/detect`. Fully offline.

Reformat a MAC address into the requested notation (colon, hyphen, Cisco dot or bare) and case (lower default or upper).

Render GitHub-Flavored Markdown to HTML. Output is sanitized by default (scripts, event handlers and javascript: URLs removed); set sanitize=false to keep raw HTML.

Replace matches of a RE2 pattern with a replacement string. Replaces all matches by default (set all=false for first-only). The replacement may reference capture groups with $1, $2 or ${name}. Returns the resulting string.

Resolve DNS records for a hostname — A, AAAA, CNAME, MX, TXT, NS, SOA, CAA or SRV. NXDOMAIN or an absent record type returns an empty record set.

Resolve a relative or absolute reference against a base URL, following RFC 3986 reference resolution (e.g. base 'https://a.com/x/y' + ref '../z' → 'https://a.com/z'). The base must itself be a valid absolute URL, else a clean 400.

Reverse-resolve an IPv4 or IPv6 address to its PTR hostname(s). No PTR record returns an empty list. This is reverse DNS (PTR) — not ENS or onchain name resolution.

Roll dice notation with a cryptographically secure RNG, optionally repeated. Returns per-die rolls, the kept dice and the total for each roll.

Convert between integers (1–3999) and Roman numerals in either direction — a digit string is encoded to Roman, a Roman numeral is decoded to an integer. Non-canonical numerals (IIII, IC) are rejected.

Round an amount to a currency's minor units (or an explicit minorUnits) with a chosen rounding mode: half-even (banker's, default), half-up, half-down, ceil, floor or trunc. Give a currency to derive minor units automatically (USD→2, JPY→0).

Sanitize untrusted HTML to a safe allowlist. Keeps common formatting and links but removes dangerous scripting tags, inline event handlers and unsafe URL schemes. Ideal for rendering user-supplied HTML.

Scan source code or configuration text for hardcoded secrets: AWS access keys, GitHub/Slack/Google tokens, Stripe and OpenAI keys, PEM private-key blocks, JWTs and high-entropy secret assignments. Placeholder values are filtered out. Each finding gives the rule, a severity (critical/high/medium/low), line and column, entropy and a masked excerpt. Offline and deterministic.

Scholarly search API - academic papers and research articles for a query, powered by OpenAlex. Returns ranked works with title, authors, year, venue, DOI, citation count and open-access status. A google-scholar-style research / scholarly articles search for agents. Data source: OpenAlex (CC0).

Assess how trustworthy a domain is using authoritative live signals, not just its spelling. Fuses RDAP registration data (domain age — the single strongest phishing signal — plus registrar, expiry and registry status flags), DNS resolvability (A/AAAA), MX presence and DNSSEC into a 0-100 risk score with a verdict (low/medium/high) and the exact reasons. Pairs with `urltrust` (lexical) for a complete picture.

Fuzzy-search countries, currencies and languages by name or code. Scope with `registry` (defaults to all). Matches are ranked exact-code/name → prefix → substring.

Live Google autocomplete suggestions for a partial query, localised by location and language. Returns ranked suggestion strings with a relevance score.

Fuzzy-search emoji by name, shortcode or keyword tag, ranked exact → prefix → substring.

Find Legal Entity Identifiers by legal name, optionally filtered to an ISO 3166-1 alpha-2 country. Returns matching entities with their LEI, verified legal name, registration status, jurisdiction and city/country - the reverse of lookup, for when you have a company name but need its authoritative LEI and legal identity.

Grow a principal by simple or compound interest over a number of years. Compound interest uses compoundsPerYear compounding periods (default 12, monthly).

Reduce a fraction to lowest terms and describe it: numerator/denominator (denominator normalized positive), sign, whether it is an integer, and a mixed-number form when improper. A zero denominator is a clean 400.

Return the singular form of an English word (handles irregulars — people → person, mice → mouse).

Solve the linear system A·x = b for a square coefficient matrix A and vector b via Gaussian elimination with partial pivoting. A non-square A or a b of the wrong length is a clean 400. A singular system is NOT an error — it returns a successful { solvable:false, solution:null }.

Encode a word with the classic Soundex algorithm (letter + 3 digits) for phonetic matching of names.

Spell text using the NATO/ICAO phonetic alphabet (A→Alfa, B→Bravo, …), including digits and spaces (→Space). Characters with no phonetic spelling are clean 400s.

Split the input into parts on each match of a RE2 pattern. An optional limit caps the number of parts (0 = unlimited). Returns the parts array and its count.

Remove all HTML tags, returning the plain text content. Entities are decoded to real characters by default (set decodeEntities=false to keep them encoded).

Student's t distribution with df degrees of freedom (> 0). `function` selects the density (pdf), cumulative distribution (cdf) or quantile / inverse-CDF.

Compute network details of a CIDR block — network & broadcast address, usable host range, total and usable address counts. Works for IPv4 and IPv6 (counts returned as strings for large IPv6 blocks). Honours RFC 3021 for /31.

Compute sunrise, sunset, solar noon and day length for a latitude/longitude (east-positive) and date, using the NOAA sunrise equation. All times are UTC ISO-8601. Locations in polar day or night return null rise/set with a `polar` flag — a successful result, not an error.

Extract the heading outline from Markdown — depth, text and a unique GitHub-style anchor slug for each heading. Optionally limit to a maximum heading depth.

Test whether a RE2 pattern matches anywhere in the input. Returns { matched }. RE2 guarantees linear-time evaluation, so this is safe against catastrophic backtracking. An invalid pattern is a clean 400; no match is a successful 200.

Transcribe text to Unicode braille (grade-1 / uncontracted English braille), with capital and number signs. Characters with no grade-1 representation are clean 400s.

Map a WGS84 coordinate at a given zoom (0..30) to its slippy-map (XYZ / Google / OSM) tile x,y and its Bing quadkey.

Compute the geographic bounding box (west/south/east/north in degrees) of a slippy-map tile z/x/y, plus its Bing quadkey.

Add or subtract from a timecode, either another timecode (b) or a raw frame count (frames), at a given frame rate. Uses frame-exact SMPTE math (including drop-frame). A result below zero is rejected as a clean 400.

Resolve the timezone for a latitude/longitude.

Convert a decimal-degree latitude/longitude to degrees-minutes-seconds, with hemisphere letters and a formatted string for each axis.

Convert a proleptic-Gregorian calendar date to its Julian Day Number (the integer count of days since JDN 0, 1 January 4713 BC in the proleptic Julian calendar).

Decode a base64 byte buffer from one character set and re-encode it into another (utf-8, utf-16le, utf-16be, latin1, ascii). Characters that the target set cannot represent, or bytes invalid for the source set, are clean 400s.

Transpose a matrix, swapping rows and columns. A ragged/empty matrix is a clean 400.

Transpose a note by a (possibly negative) number of semitones. Result is spelled chromatically with sharps, or flats when `preferFlats` is true.

Apply a Unicode normalisation form (NFC, NFD, NFKC or NFKD) to a string. Reports whether the text changed and its length in UTF-16 code units and code points.

Convert decimal lat/lon to a UTM coordinate (zone, latitude band, hemisphere, easting, northing) on WGS84, including the Norway/Svalbard zone exceptions. Defined for latitudes -80..84.

Validate a Czech VAT (DIČ) number for a legal entity against the official register. Validity means the subject exists and is live in the VAT (DPH) register.

Validate a Finnish VAT (ALV) number against the official VAT register. Validity means the entity exists and holds a live registration.

Validate a French VAT (TVA) number against the official company register. Validity means the SIREN exists and the company is administratively active.

Validate a Latvian VAT (PVN) number against the official VAT-payers register. Validity means the number has a currently active entry in the register.

Validate a Norwegian VAT (MVA) number against the official business register. Validity means the entity exists and is entered in the VAT register.

Validate a Polish VAT (NIP) number against the official VAT register. Validity means the taxpayer's VAT status is active (Czynny).

Validate a payment-card number: Luhn check-digit test plus network detection and length check. A syntactically valid number that fails Luhn returns `valid:false` (a 200, not an error).

Check whether a postal / ZIP code is well-formed for a country (ISO 3166-1 alpha-2). A malformed code is a successful result with `valid:false`, not an error — only an unsupported country or empty code is a 400.

Validate a JSON value against a JSON Schema (draft-07, with the standard string formats enabled). An invalid schema is rejected as a clean 400 before payment; a valid schema that the data simply fails to satisfy is a successful result with valid=false and a list of errors (path, keyword, message).

Parse a document (json, json5, jsonc, yaml, xml, csv, tsv, toml, ndjson, ini, env, properties, plist or querystring) and validate it against a JSON Schema. Returns whether it is valid and a list of precise, path-tagged errors. Malformed input or an invalid schema is rejected before payment.

Validate an email address — syntax, MX/A deliverability lookup, and disposable / role-based address detection. No mailbox probing.

Compute the WCAG 2.1 contrast ratio between two colors and whether it passes AA/AAA.

Web search API - live Google (default) or Bing SERP for a query: search engine results returned as ranked organic hits with title, URL, domain and snippet, localised by location and language. A google search api / serp api for agents: search the web, live web search for grounding, RAG and web lookup.

Detect the technology stack a website runs on. Given a domain, returns the software and services it uses (CDN, analytics, frameworks, CMS, payment, hosting, security) grouped by category, plus a domain rank, last-seen date, country and public social links. Unknown domain returns found=false.

Wikipedia summary API - the plain-language summary of the best-matching Wikipedia article for a topic or query, in any language. Returns the article title, short description, summary extract, thumbnail image and canonical URL. A wikipedia / encyclopedia lookup for agents grounding answers on a topic. No match returns found:false. Data source: Wikipedia (CC BY-SA 4.0).

Live YouTube video-search results for a query, localised by location and language. Returns ranked videos with title, URL, video id, channel, view count, publish date and duration.