@terminal3/t3n-sdk 5.14.0 → 5.15.0

package/dist/scripts/sign-trust-manifest.mjs

@@ -0,0 +1,3903 @@
+#!/usr/bin/env node
+/**
+ * Checks if something is Uint8Array. Be careful: nodejs Buffer will return true.
+ * @param a - value to test
+ * @returns `true` when the value is a Uint8Array-compatible view.
+ * @example
+ * Check whether a value is a Uint8Array-compatible view.
+ * ```ts
+ * isBytes(new Uint8Array([1, 2, 3]));
+ * ```
+ */
+function isBytes$1(a) {
+    // Plain `instanceof Uint8Array` is too strict for some Buffer / proxy / cross-realm cases.
+    // The fallback still requires a real ArrayBuffer view, so plain
+    // JSON-deserialized `{ constructor: ... }` spoofing is rejected, and
+    // `BYTES_PER_ELEMENT === 1` keeps the fallback on byte-oriented views.
+    return (a instanceof Uint8Array ||
+        (ArrayBuffer.isView(a) &&
+            a.constructor.name === 'Uint8Array' &&
+            'BYTES_PER_ELEMENT' in a &&
+            a.BYTES_PER_ELEMENT === 1));
+}
+/**
+ * Asserts something is a non-negative integer.
+ * @param n - number to validate
+ * @param title - label included in thrown errors
+ * @throws On wrong argument types. {@link TypeError}
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Validate a non-negative integer option.
+ * ```ts
+ * anumber(32, 'length');
+ * ```
+ */
+function anumber$1(n, title = '') {
+    if (typeof n !== 'number') {
+        const prefix = title && `"${title}" `;
+        throw new TypeError(`${prefix}expected number, got ${typeof n}`);
+    }
+    if (!Number.isSafeInteger(n) || n < 0) {
+        const prefix = title && `"${title}" `;
+        throw new RangeError(`${prefix}expected integer >= 0, got ${n}`);
+    }
+}
+/**
+ * Asserts something is Uint8Array.
+ * @param value - value to validate
+ * @param length - optional exact length constraint
+ * @param title - label included in thrown errors
+ * @returns The validated byte array.
+ * @throws On wrong argument types. {@link TypeError}
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Validate that a value is a byte array.
+ * ```ts
+ * abytes(new Uint8Array([1, 2, 3]));
+ * ```
+ */
+function abytes$1(value, length, title = '') {
+    const bytes = isBytes$1(value);
+    const len = value?.length;
+    const needsLen = length !== undefined;
+    if (!bytes || (needsLen && len !== length)) {
+        const prefix = title && `"${title}" `;
+        const ofLen = needsLen ? ` of length ${length}` : '';
+        const got = bytes ? `length=${len}` : `type=${typeof value}`;
+        const message = prefix + 'expected Uint8Array' + ofLen + ', got ' + got;
+        if (!bytes)
+            throw new TypeError(message);
+        throw new RangeError(message);
+    }
+    return value;
+}
+/**
+ * Asserts something is a wrapped hash constructor.
+ * @param h - hash constructor to validate
+ * @throws On wrong argument types or invalid hash wrapper shape. {@link TypeError}
+ * @throws On invalid hash metadata ranges or values. {@link RangeError}
+ * @throws If the hash metadata allows empty outputs or block sizes. {@link Error}
+ * @example
+ * Validate a callable hash wrapper.
+ * ```ts
+ * import { ahash } from '@noble/hashes/utils.js';
+ * import { sha256 } from '@noble/hashes/sha2.js';
+ * ahash(sha256);
+ * ```
+ */
+function ahash(h) {
+    if (typeof h !== 'function' || typeof h.create !== 'function')
+        throw new TypeError('Hash must wrapped by utils.createHasher');
+    anumber$1(h.outputLen);
+    anumber$1(h.blockLen);
+    // HMAC and KDF callers treat these as real byte lengths; allowing zero lets fake wrappers pass
+    // validation and can produce empty outputs instead of failing fast.
+    if (h.outputLen < 1)
+        throw new Error('"outputLen" must be >= 1');
+    if (h.blockLen < 1)
+        throw new Error('"blockLen" must be >= 1');
+}
+/**
+ * Asserts a hash instance has not been destroyed or finished.
+ * @param instance - hash instance to validate
+ * @param checkFinished - whether to reject finalized instances
+ * @throws If the hash instance has already been destroyed or finalized. {@link Error}
+ * @example
+ * Validate that a hash instance is still usable.
+ * ```ts
+ * import { aexists } from '@noble/hashes/utils.js';
+ * import { sha256 } from '@noble/hashes/sha2.js';
+ * const hash = sha256.create();
+ * aexists(hash);
+ * ```
+ */
+function aexists(instance, checkFinished = true) {
+    if (instance.destroyed)
+        throw new Error('Hash instance has been destroyed');
+    if (checkFinished && instance.finished)
+        throw new Error('Hash#digest() has already been called');
+}
+/**
+ * Asserts output is a sufficiently-sized byte array.
+ * @param out - destination buffer
+ * @param instance - hash instance providing output length
+ * Oversized buffers are allowed; downstream code only promises to fill the first `outputLen` bytes.
+ * @throws On wrong argument types. {@link TypeError}
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Validate a caller-provided digest buffer.
+ * ```ts
+ * import { aoutput } from '@noble/hashes/utils.js';
+ * import { sha256 } from '@noble/hashes/sha2.js';
+ * const hash = sha256.create();
+ * aoutput(new Uint8Array(hash.outputLen), hash);
+ * ```
+ */
+function aoutput(out, instance) {
+    abytes$1(out, undefined, 'digestInto() output');
+    const min = instance.outputLen;
+    if (out.length < min) {
+        throw new RangeError('"digestInto() output" expected to be of length >=' + min);
+    }
+}
+/**
+ * Zeroizes typed arrays in place. Warning: JS provides no guarantees.
+ * @param arrays - arrays to overwrite with zeros
+ * @example
+ * Zeroize sensitive buffers in place.
+ * ```ts
+ * clean(new Uint8Array([1, 2, 3]));
+ * ```
+ */
+function clean(...arrays) {
+    for (let i = 0; i < arrays.length; i++) {
+        arrays[i].fill(0);
+    }
+}
+/**
+ * Creates a DataView for byte-level manipulation.
+ * @param arr - source typed array
+ * @returns DataView over the same buffer region.
+ * @example
+ * Create a DataView over an existing buffer.
+ * ```ts
+ * createView(new Uint8Array(4));
+ * ```
+ */
+function createView(arr) {
+    return new DataView(arr.buffer, arr.byteOffset, arr.byteLength);
+}
+/**
+ * Rotate-right operation for uint32 values.
+ * @param word - source word
+ * @param shift - shift amount in bits
+ * @returns Rotated word.
+ * @example
+ * Rotate a 32-bit word to the right.
+ * ```ts
+ * rotr(0x12345678, 8);
+ * ```
+ */
+function rotr(word, shift) {
+    return (word << (32 - shift)) | (word >>> shift);
+}
+// Built-in hex conversion https://caniuse.com/mdn-javascript_builtins_uint8array_fromhex
+const hasHexBuiltin = /* @__PURE__ */ (() => 
+// @ts-ignore
+typeof Uint8Array.from([]).toHex === 'function' && typeof Uint8Array.fromHex === 'function')();
+// Array where index 0xf0 (240) is mapped to string 'f0'
+const hexes = /* @__PURE__ */ Array.from({ length: 256 }, (_, i) => i.toString(16).padStart(2, '0'));
+/**
+ * Convert byte array to hex string.
+ * Uses the built-in function when available and assumes it matches the tested
+ * fallback semantics.
+ * @param bytes - bytes to encode
+ * @returns Lowercase hexadecimal string.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Convert bytes to lowercase hexadecimal.
+ * ```ts
+ * bytesToHex(Uint8Array.from([0xca, 0xfe, 0x01, 0x23])); // 'cafe0123'
+ * ```
+ */
+function bytesToHex$1(bytes) {
+    abytes$1(bytes);
+    // @ts-ignore
+    if (hasHexBuiltin)
+        return bytes.toHex();
+    // pre-caching improves the speed 6x
+    let hex = '';
+    for (let i = 0; i < bytes.length; i++) {
+        hex += hexes[bytes[i]];
+    }
+    return hex;
+}
+// We use optimized technique to convert hex string to byte array
+const asciis = { _0: 48, _9: 57, A: 65, F: 70, a: 97, f: 102 };
+function asciiToBase16(ch) {
+    if (ch >= asciis._0 && ch <= asciis._9)
+        return ch - asciis._0; // '2' => 50-48
+    if (ch >= asciis.A && ch <= asciis.F)
+        return ch - (asciis.A - 10); // 'B' => 66-(65-10)
+    if (ch >= asciis.a && ch <= asciis.f)
+        return ch - (asciis.a - 10); // 'b' => 98-(97-10)
+    return;
+}
+/**
+ * Convert hex string to byte array. Uses built-in function, when available.
+ * @param hex - hexadecimal string to decode
+ * @returns Decoded bytes.
+ * @throws On wrong argument types. {@link TypeError}
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Decode lowercase hexadecimal into bytes.
+ * ```ts
+ * hexToBytes('cafe0123'); // Uint8Array.from([0xca, 0xfe, 0x01, 0x23])
+ * ```
+ */
+function hexToBytes$1(hex) {
+    if (typeof hex !== 'string')
+        throw new TypeError('hex string expected, got ' + typeof hex);
+    if (hasHexBuiltin) {
+        try {
+            return Uint8Array.fromHex(hex);
+        }
+        catch (error) {
+            if (error instanceof SyntaxError)
+                throw new RangeError(error.message);
+            throw error;
+        }
+    }
+    const hl = hex.length;
+    const al = hl / 2;
+    if (hl % 2)
+        throw new RangeError('hex string expected, got unpadded hex of length ' + hl);
+    const array = new Uint8Array(al);
+    for (let ai = 0, hi = 0; ai < al; ai++, hi += 2) {
+        const n1 = asciiToBase16(hex.charCodeAt(hi));
+        const n2 = asciiToBase16(hex.charCodeAt(hi + 1));
+        if (n1 === undefined || n2 === undefined) {
+            const char = hex[hi] + hex[hi + 1];
+            throw new RangeError('hex string expected, got non-hex character "' + char + '" at index ' + hi);
+        }
+        array[ai] = n1 * 16 + n2; // multiply first octet, e.g. 'a3' => 10*16+3 => 160 + 3 => 163
+    }
+    return array;
+}
+/**
+ * Copies several Uint8Arrays into one.
+ * @param arrays - arrays to concatenate
+ * @returns Concatenated byte array.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Concatenate multiple byte arrays.
+ * ```ts
+ * concatBytes(new Uint8Array([1]), new Uint8Array([2]));
+ * ```
+ */
+function concatBytes$1(...arrays) {
+    let sum = 0;
+    for (let i = 0; i < arrays.length; i++) {
+        const a = arrays[i];
+        abytes$1(a);
+        sum += a.length;
+    }
+    const res = new Uint8Array(sum);
+    for (let i = 0, pad = 0; i < arrays.length; i++) {
+        const a = arrays[i];
+        res.set(a, pad);
+        pad += a.length;
+    }
+    return res;
+}
+/**
+ * Creates a callable hash function from a stateful class constructor.
+ * @param hashCons - hash constructor or factory
+ * @param info - optional metadata such as DER OID
+ * @returns Frozen callable hash wrapper with `.create()`.
+ *   Wrapper construction eagerly calls `hashCons(undefined)` once to read
+ *   `outputLen` / `blockLen`, so constructor side effects happen at module
+ *   init time.
+ * @example
+ * Wrap a stateful hash constructor into a callable helper.
+ * ```ts
+ * import { createHasher } from '@noble/hashes/utils.js';
+ * import { sha256 } from '@noble/hashes/sha2.js';
+ * const wrapped = createHasher(sha256.create, { oid: sha256.oid });
+ * wrapped(new Uint8Array([1]));
+ * ```
+ */
+function createHasher(hashCons, info = {}) {
+    const hashC = (msg, opts) => hashCons(opts)
+        .update(msg)
+        .digest();
+    const tmp = hashCons(undefined);
+    hashC.outputLen = tmp.outputLen;
+    hashC.blockLen = tmp.blockLen;
+    hashC.canXOF = tmp.canXOF;
+    hashC.create = (opts) => hashCons(opts);
+    Object.assign(hashC, info);
+    return Object.freeze(hashC);
+}
+/**
+ * Cryptographically secure PRNG backed by `crypto.getRandomValues`.
+ * @param bytesLength - number of random bytes to generate
+ * @returns Random bytes.
+ * The platform `getRandomValues()` implementation still defines any
+ * single-call length cap, and this helper rejects oversize requests
+ * with a stable library `RangeError` instead of host-specific errors.
+ * @throws On wrong argument types. {@link TypeError}
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @throws If the current runtime does not provide `crypto.getRandomValues`. {@link Error}
+ * @example
+ * Generate a fresh random key or nonce.
+ * ```ts
+ * const key = randomBytes(16);
+ * ```
+ */
+function randomBytes$1(bytesLength = 32) {
+    // Match the repo's other length-taking helpers instead of relying on Uint8Array coercion.
+    anumber$1(bytesLength, 'bytesLength');
+    const cr = typeof globalThis === 'object' ? globalThis.crypto : null;
+    if (typeof cr?.getRandomValues !== 'function')
+        throw new Error('crypto.getRandomValues must be defined');
+    // Web Cryptography API Level 2 §10.1.1:
+    // if `byteLength > 65536`, throw `QuotaExceededError`.
+    // Keep the guard explicit so callers can see the quota in code
+    // instead of discovering it by reading the spec or host errors.
+    // This wrapper surfaces the same quota as a stable library RangeError.
+    if (bytesLength > 65536)
+        throw new RangeError(`"bytesLength" expected <= 65536, got ${bytesLength}`);
+    return cr.getRandomValues(new Uint8Array(bytesLength));
+}
+/**
+ * Creates OID metadata for NIST hashes with prefix `06 09 60 86 48 01 65 03 04 02`.
+ * @param suffix - final OID byte for the selected hash.
+ *   The helper accepts any byte even though only the documented NIST hash
+ *   suffixes are meaningful downstream.
+ * @returns Object containing the DER-encoded OID.
+ * @example
+ * Build OID metadata for a NIST hash.
+ * ```ts
+ * oidNist(0x01);
+ * ```
+ */
+const oidNist = (suffix) => ({
+    // Current NIST hashAlgs suffixes used here fit in one DER subidentifier octet.
+    // Larger suffix values would need base-128 OID encoding and a different length byte.
+    oid: Uint8Array.from([0x06, 0x09, 0x60, 0x86, 0x48, 0x01, 0x65, 0x03, 0x04, 0x02, suffix]),
+});
+
+/**
+ * Internal Merkle-Damgard hash utils.
+ * @module
+ */
+/**
+ * Shared 32-bit conditional boolean primitive reused by SHA-256, SHA-1, and MD5 `F`.
+ * Returns bits from `b` when `a` is set, otherwise from `c`.
+ * The XOR form is equivalent to MD5's `F(X,Y,Z) = XY v not(X)Z` because the masked terms never
+ * set the same bit.
+ * @param a - selector word
+ * @param b - word chosen when selector bit is set
+ * @param c - word chosen when selector bit is clear
+ * @returns Mixed 32-bit word.
+ * @example
+ * Combine three words with the shared 32-bit choice primitive.
+ * ```ts
+ * Chi(0xffffffff, 0x12345678, 0x87654321);
+ * ```
+ */
+function Chi(a, b, c) {
+    return (a & b) ^ (~a & c);
+}
+/**
+ * Shared 32-bit majority primitive reused by SHA-256 and SHA-1.
+ * Returns bits shared by at least two inputs.
+ * @param a - first input word
+ * @param b - second input word
+ * @param c - third input word
+ * @returns Mixed 32-bit word.
+ * @example
+ * Combine three words with the shared 32-bit majority primitive.
+ * ```ts
+ * Maj(0xffffffff, 0x12345678, 0x87654321);
+ * ```
+ */
+function Maj(a, b, c) {
+    return (a & b) ^ (a & c) ^ (b & c);
+}
+/**
+ * Merkle-Damgard hash construction base class.
+ * Could be used to create MD5, RIPEMD, SHA1, SHA2.
+ * Accepts only byte-aligned `Uint8Array` input, even when the underlying spec describes bit
+ * strings with partial-byte tails.
+ * @param blockLen - internal block size in bytes
+ * @param outputLen - digest size in bytes
+ * @param padOffset - trailing length field size in bytes
+ * @param isLE - whether length and state words are encoded in little-endian
+ * @example
+ * Use a concrete subclass to get the shared Merkle-Damgard update/digest flow.
+ * ```ts
+ * import { _SHA1 } from '@noble/hashes/legacy.js';
+ * const hash = new _SHA1();
+ * hash.update(new Uint8Array([97, 98, 99]));
+ * hash.digest();
+ * ```
+ */
+class HashMD {
+    blockLen;
+    outputLen;
+    canXOF = false;
+    padOffset;
+    isLE;
+    // For partial updates less than block size
+    buffer;
+    view;
+    finished = false;
+    length = 0;
+    pos = 0;
+    destroyed = false;
+    constructor(blockLen, outputLen, padOffset, isLE) {
+        this.blockLen = blockLen;
+        this.outputLen = outputLen;
+        this.padOffset = padOffset;
+        this.isLE = isLE;
+        this.buffer = new Uint8Array(blockLen);
+        this.view = createView(this.buffer);
+    }
+    update(data) {
+        aexists(this);
+        abytes$1(data);
+        const { view, buffer, blockLen } = this;
+        const len = data.length;
+        for (let pos = 0; pos < len;) {
+            const take = Math.min(blockLen - this.pos, len - pos);
+            // Fast path only when there is no buffered partial block: `take === blockLen` implies
+            // `this.pos === 0`, so we can process full blocks directly from the input view.
+            if (take === blockLen) {
+                const dataView = createView(data);
+                for (; blockLen <= len - pos; pos += blockLen)
+                    this.process(dataView, pos);
+                continue;
+            }
+            buffer.set(data.subarray(pos, pos + take), this.pos);
+            this.pos += take;
+            pos += take;
+            if (this.pos === blockLen) {
+                this.process(view, 0);
+                this.pos = 0;
+            }
+        }
+        this.length += data.length;
+        this.roundClean();
+        return this;
+    }
+    digestInto(out) {
+        aexists(this);
+        aoutput(out, this);
+        this.finished = true;
+        // Padding
+        // We can avoid allocation of buffer for padding completely if it
+        // was previously not allocated here. But it won't change performance.
+        const { buffer, view, blockLen, isLE } = this;
+        let { pos } = this;
+        // append the bit '1' to the message
+        buffer[pos++] = 0b10000000;
+        clean(this.buffer.subarray(pos));
+        // we have less than padOffset left in buffer, so we cannot put length in
+        // current block, need process it and pad again
+        if (this.padOffset > blockLen - pos) {
+            this.process(view, 0);
+            pos = 0;
+        }
+        // Pad until full block byte with zeros
+        for (let i = pos; i < blockLen; i++)
+            buffer[i] = 0;
+        // `padOffset` reserves the whole length field. For SHA-384/512 the high 64 bits stay zero from
+        // the padding fill above, and JS will overflow before user input can make that half non-zero.
+        // So we only need to write the low 64 bits here.
+        view.setBigUint64(blockLen - 8, BigInt(this.length * 8), isLE);
+        this.process(view, 0);
+        const oview = createView(out);
+        const len = this.outputLen;
+        // NOTE: we do division by 4 later, which must be fused in single op with modulo by JIT
+        if (len % 4)
+            throw new Error('_sha2: outputLen must be aligned to 32bit');
+        const outLen = len / 4;
+        const state = this.get();
+        if (outLen > state.length)
+            throw new Error('_sha2: outputLen bigger than state');
+        for (let i = 0; i < outLen; i++)
+            oview.setUint32(4 * i, state[i], isLE);
+    }
+    digest() {
+        const { buffer, outputLen } = this;
+        this.digestInto(buffer);
+        // Copy before destroy(): subclasses wipe `buffer` during cleanup, but `digest()` must return
+        // fresh bytes to the caller.
+        const res = buffer.slice(0, outputLen);
+        this.destroy();
+        return res;
+    }
+    _cloneInto(to) {
+        to ||= new this.constructor();
+        to.set(...this.get());
+        const { blockLen, buffer, length, finished, destroyed, pos } = this;
+        to.destroyed = destroyed;
+        to.finished = finished;
+        to.length = length;
+        to.pos = pos;
+        // Only partial-block bytes need copying: when `length % blockLen === 0`, `pos === 0` and
+        // later `update()` / `digestInto()` overwrite `to.buffer` from the start before reading it.
+        if (length % blockLen)
+            to.buffer.set(buffer);
+        return to;
+    }
+    clone() {
+        return this._cloneInto();
+    }
+}
+/**
+ * Initial SHA-2 state: fractional parts of square roots of first 16 primes 2..53.
+ * Check out `test/misc/sha2-gen-iv.js` for recomputation guide.
+ */
+/** Initial SHA256 state from RFC 6234 §6.1: the first 32 bits of the fractional parts of the
+ * square roots of the first eight prime numbers. Exported as a shared table; callers must treat
+ * it as read-only because constructors copy words from it by index. */
+const SHA256_IV = /* @__PURE__ */ Uint32Array.from([
+    0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19,
+]);
+
+/**
+ * SHA2 hash function. A.k.a. sha256, sha384, sha512, sha512_224, sha512_256.
+ * SHA256 is the fastest hash implementable in JS, even faster than Blake3.
+ * Check out {@link https://www.rfc-editor.org/rfc/rfc4634 | RFC 4634} and
+ * {@link https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.180-4.pdf | FIPS 180-4}.
+ * @module
+ */
+/**
+ * SHA-224 / SHA-256 round constants from RFC 6234 §5.1: the first 32 bits
+ * of the cube roots of the first 64 primes (2..311).
+ */
+// prettier-ignore
+const SHA256_K = /* @__PURE__ */ Uint32Array.from([
+    0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
+    0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
+    0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
+    0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
+    0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
+    0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
+    0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
+    0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
+]);
+/** Reusable SHA-224 / SHA-256 message schedule buffer `W_t` from RFC 6234 §6.2 step 1. */
+const SHA256_W = /* @__PURE__ */ new Uint32Array(64);
+/** Internal SHA-224 / SHA-256 compression engine from RFC 6234 §6.2. */
+class SHA2_32B extends HashMD {
+    constructor(outputLen) {
+        super(64, outputLen, 8, false);
+    }
+    get() {
+        const { A, B, C, D, E, F, G, H } = this;
+        return [A, B, C, D, E, F, G, H];
+    }
+    // prettier-ignore
+    set(A, B, C, D, E, F, G, H) {
+        this.A = A | 0;
+        this.B = B | 0;
+        this.C = C | 0;
+        this.D = D | 0;
+        this.E = E | 0;
+        this.F = F | 0;
+        this.G = G | 0;
+        this.H = H | 0;
+    }
+    process(view, offset) {
+        // Extend the first 16 words into the remaining 48 words w[16..63] of the message schedule array
+        for (let i = 0; i < 16; i++, offset += 4)
+            SHA256_W[i] = view.getUint32(offset, false);
+        for (let i = 16; i < 64; i++) {
+            const W15 = SHA256_W[i - 15];
+            const W2 = SHA256_W[i - 2];
+            const s0 = rotr(W15, 7) ^ rotr(W15, 18) ^ (W15 >>> 3);
+            const s1 = rotr(W2, 17) ^ rotr(W2, 19) ^ (W2 >>> 10);
+            SHA256_W[i] = (s1 + SHA256_W[i - 7] + s0 + SHA256_W[i - 16]) | 0;
+        }
+        // Compression function main loop, 64 rounds
+        let { A, B, C, D, E, F, G, H } = this;
+        for (let i = 0; i < 64; i++) {
+            const sigma1 = rotr(E, 6) ^ rotr(E, 11) ^ rotr(E, 25);
+            const T1 = (H + sigma1 + Chi(E, F, G) + SHA256_K[i] + SHA256_W[i]) | 0;
+            const sigma0 = rotr(A, 2) ^ rotr(A, 13) ^ rotr(A, 22);
+            const T2 = (sigma0 + Maj(A, B, C)) | 0;
+            H = G;
+            G = F;
+            F = E;
+            E = (D + T1) | 0;
+            D = C;
+            C = B;
+            B = A;
+            A = (T1 + T2) | 0;
+        }
+        // Add the compressed chunk to the current hash value
+        A = (A + this.A) | 0;
+        B = (B + this.B) | 0;
+        C = (C + this.C) | 0;
+        D = (D + this.D) | 0;
+        E = (E + this.E) | 0;
+        F = (F + this.F) | 0;
+        G = (G + this.G) | 0;
+        H = (H + this.H) | 0;
+        this.set(A, B, C, D, E, F, G, H);
+    }
+    roundClean() {
+        clean(SHA256_W);
+    }
+    destroy() {
+        // HashMD callers route post-destroy usability through `destroyed`; zeroizing alone still leaves
+        // update()/digest() callable on reused instances.
+        this.destroyed = true;
+        this.set(0, 0, 0, 0, 0, 0, 0, 0);
+        clean(this.buffer);
+    }
+}
+/** Internal SHA-256 hash class grounded in RFC 6234 §6.2. */
+class _SHA256 extends SHA2_32B {
+    // We cannot use array here since array allows indexing by variable
+    // which means optimizer/compiler cannot use registers.
+    A = SHA256_IV[0] | 0;
+    B = SHA256_IV[1] | 0;
+    C = SHA256_IV[2] | 0;
+    D = SHA256_IV[3] | 0;
+    E = SHA256_IV[4] | 0;
+    F = SHA256_IV[5] | 0;
+    G = SHA256_IV[6] | 0;
+    H = SHA256_IV[7] | 0;
+    constructor() {
+        super(32);
+    }
+}
+/**
+ * SHA2-256 hash function from RFC 4634. In JS it's the fastest: even faster than Blake3. Some info:
+ *
+ * - Trying 2^128 hashes would get 50% chance of collision, using birthday attack.
+ * - BTC network is doing 2^70 hashes/sec (2^95 hashes/year) as per 2025.
+ * - Each sha256 hash is executing 2^18 bit operations.
+ * - Good 2024 ASICs can do 200Th/sec with 3500 watts of power, corresponding to 2^36 hashes/joule.
+ * @param msg - message bytes to hash
+ * @returns Digest bytes.
+ * @example
+ * Hash a message with SHA2-256.
+ * ```ts
+ * sha256(new Uint8Array([97, 98, 99]));
+ * ```
+ */
+const sha256 = /* @__PURE__ */ createHasher(() => new _SHA256(), 
+/* @__PURE__ */ oidNist(0x01));
+
+/**
+ * Hex, bytes and number utilities.
+ * @module
+ */
+/*! noble-curves - MIT License (c) 2022 Paul Miller (paulmillr.com) */
+/**
+ * Validates that a value is a byte array.
+ * @param value - Value to validate.
+ * @param length - Optional exact byte length.
+ * @param title - Optional field name.
+ * @returns Original byte array.
+ * @example
+ * Reject non-byte input before passing data into curve code.
+ *
+ * ```ts
+ * abytes(new Uint8Array(1));
+ * ```
+ */
+const abytes = (value, length, title) => abytes$1(value, length, title);
+/**
+ * Validates that a value is a non-negative safe integer.
+ * @param n - Value to validate.
+ * @param title - Optional field name.
+ * @example
+ * Validate a numeric length before allocating buffers.
+ *
+ * ```ts
+ * anumber(1);
+ * ```
+ */
+const anumber = anumber$1;
+/**
+ * Encodes bytes as lowercase hex.
+ * @param bytes - Bytes to encode.
+ * @returns Lowercase hex string.
+ * @example
+ * Serialize bytes as hex for logging or fixtures.
+ *
+ * ```ts
+ * bytesToHex(Uint8Array.of(1, 2, 3));
+ * ```
+ */
+const bytesToHex = bytesToHex$1;
+/**
+ * Concatenates byte arrays.
+ * @param arrays - Byte arrays to join.
+ * @returns Concatenated bytes.
+ * @example
+ * Join domain-separated chunks into one buffer.
+ *
+ * ```ts
+ * concatBytes(Uint8Array.of(1), Uint8Array.of(2));
+ * ```
+ */
+const concatBytes = (...arrays) => concatBytes$1(...arrays);
+/**
+ * Decodes lowercase or uppercase hex into bytes.
+ * @param hex - Hex string to decode.
+ * @returns Decoded bytes.
+ * @example
+ * Parse fixture hex into bytes before hashing.
+ *
+ * ```ts
+ * hexToBytes('0102');
+ * ```
+ */
+const hexToBytes = (hex) => hexToBytes$1(hex);
+/**
+ * Checks whether a value is a Uint8Array.
+ * @param a - Value to inspect.
+ * @returns `true` when `a` is a Uint8Array.
+ * @example
+ * Branch on byte input before decoding it.
+ *
+ * ```ts
+ * isBytes(new Uint8Array(1));
+ * ```
+ */
+const isBytes = isBytes$1;
+/**
+ * Reads random bytes from the platform CSPRNG.
+ * @param bytesLength - Number of random bytes to read.
+ * @returns Fresh random bytes.
+ * @example
+ * Generate a random seed for a keypair.
+ *
+ * ```ts
+ * randomBytes(2);
+ * ```
+ */
+const randomBytes = (bytesLength) => randomBytes$1(bytesLength);
+const _0n$3 = /* @__PURE__ */ BigInt(0);
+const _1n$3 = /* @__PURE__ */ BigInt(1);
+/**
+ * Validates that a flag is boolean.
+ * @param value - Value to validate.
+ * @param title - Optional field name.
+ * @returns Original value.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Reject non-boolean option flags early.
+ *
+ * ```ts
+ * abool(true);
+ * ```
+ */
+function abool(value, title = '') {
+    if (typeof value !== 'boolean') {
+        const prefix = title && `"${title}" `;
+        throw new TypeError(prefix + 'expected boolean, got type=' + typeof value);
+    }
+    return value;
+}
+/**
+ * Validates that a value is a non-negative bigint or safe integer.
+ * @param n - Value to validate.
+ * @returns The same validated value.
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Validate one integer-like value before serializing it.
+ *
+ * ```ts
+ * abignumber(1n);
+ * ```
+ */
+function abignumber(n) {
+    if (typeof n === 'bigint') {
+        if (!isPosBig(n))
+            throw new RangeError('positive bigint expected, got ' + n);
+    }
+    else
+        anumber(n);
+    return n;
+}
+/**
+ * Validates that a value is a safe integer.
+ * @param value - Integer to validate.
+ * @param title - Optional field name.
+ * @throws On wrong argument types. {@link TypeError}
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Validate a window size before scalar arithmetic uses it.
+ *
+ * ```ts
+ * asafenumber(1);
+ * ```
+ */
+function asafenumber(value, title = '') {
+    if (typeof value !== 'number') {
+        const prefix = title && `"${title}" `;
+        throw new TypeError(prefix + 'expected number, got type=' + typeof value);
+    }
+    if (!Number.isSafeInteger(value)) {
+        const prefix = title && `"${title}" `;
+        throw new RangeError(prefix + 'expected safe integer, got ' + value);
+    }
+}
+/**
+ * Encodes a bigint into even-length big-endian hex.
+ * The historical "unpadded" name only means "no fixed-width field padding"; odd-length hex still
+ * gets one leading zero nibble so the result always represents whole bytes.
+ * @param num - Number to encode.
+ * @returns Big-endian hex string.
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Encode a scalar into hex without a `0x` prefix.
+ *
+ * ```ts
+ * numberToHexUnpadded(255n);
+ * ```
+ */
+function numberToHexUnpadded(num) {
+    const hex = abignumber(num).toString(16);
+    return hex.length & 1 ? '0' + hex : hex;
+}
+/**
+ * Parses a big-endian hex string into bigint.
+ * Accepts odd-length hex through the native `BigInt('0x' + hex)` parser and currently surfaces the
+ * same native `SyntaxError` for malformed hex instead of wrapping it in a library-specific error.
+ * @param hex - Hex string without `0x`.
+ * @returns Parsed bigint value.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Parse a scalar from fixture hex.
+ *
+ * ```ts
+ * hexToNumber('ff');
+ * ```
+ */
+function hexToNumber(hex) {
+    if (typeof hex !== 'string')
+        throw new TypeError('hex string expected, got ' + typeof hex);
+    return hex === '' ? _0n$3 : BigInt('0x' + hex); // Big Endian
+}
+// BE: Big Endian, LE: Little Endian
+/**
+ * Parses big-endian bytes into bigint.
+ * @param bytes - Bytes in big-endian order.
+ * @returns Parsed bigint value.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Read a scalar encoded in network byte order.
+ *
+ * ```ts
+ * bytesToNumberBE(Uint8Array.of(1, 0));
+ * ```
+ */
+function bytesToNumberBE(bytes) {
+    return hexToNumber(bytesToHex$1(bytes));
+}
+/**
+ * Parses little-endian bytes into bigint.
+ * @param bytes - Bytes in little-endian order.
+ * @returns Parsed bigint value.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Read a scalar encoded in little-endian form.
+ *
+ * ```ts
+ * bytesToNumberLE(Uint8Array.of(1, 0));
+ * ```
+ */
+function bytesToNumberLE(bytes) {
+    return hexToNumber(bytesToHex$1(copyBytes(abytes$1(bytes)).reverse()));
+}
+/**
+ * Encodes a bigint into fixed-length big-endian bytes.
+ * @param n - Number to encode.
+ * @param len - Output length in bytes. Must be greater than zero.
+ * @returns Big-endian byte array.
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Serialize a scalar into a 32-byte field element.
+ *
+ * ```ts
+ * numberToBytesBE(255n, 2);
+ * ```
+ */
+function numberToBytesBE(n, len) {
+    anumber$1(len);
+    if (len === 0)
+        throw new RangeError('zero length');
+    n = abignumber(n);
+    const hex = n.toString(16);
+    // Detect overflow before hex parsing so oversized values don't leak the shared odd-hex error.
+    if (hex.length > len * 2)
+        throw new RangeError('number too large');
+    return hexToBytes$1(hex.padStart(len * 2, '0'));
+}
+/**
+ * Encodes a bigint into fixed-length little-endian bytes.
+ * @param n - Number to encode.
+ * @param len - Output length in bytes.
+ * @returns Little-endian byte array.
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Serialize a scalar for little-endian protocols.
+ *
+ * ```ts
+ * numberToBytesLE(255n, 2);
+ * ```
+ */
+function numberToBytesLE(n, len) {
+    return numberToBytesBE(n, len).reverse();
+}
+/**
+ * Copies Uint8Array. We can't use u8a.slice(), because u8a can be Buffer,
+ * and Buffer#slice creates mutable copy. Never use Buffers!
+ * @param bytes - Bytes to copy.
+ * @returns Detached copy.
+ * @example
+ * Make an isolated copy before mutating serialized bytes.
+ *
+ * ```ts
+ * copyBytes(Uint8Array.of(1, 2, 3));
+ * ```
+ */
+function copyBytes(bytes) {
+    // `Uint8Array.from(...)` would also accept arrays / other typed arrays. Keep this helper strict
+    // because callers use it at byte-validation boundaries before mutating the detached copy.
+    return Uint8Array.from(abytes(bytes));
+}
+// Historical name: this accepts non-negative bigints, including zero.
+const isPosBig = (n) => typeof n === 'bigint' && _0n$3 <= n;
+/**
+ * Checks whether a bigint lies inside a half-open range.
+ * @param n - Candidate value.
+ * @param min - Inclusive lower bound.
+ * @param max - Exclusive upper bound.
+ * @returns `true` when the value is inside the range.
+ * @example
+ * Check whether a candidate scalar fits the field order.
+ *
+ * ```ts
+ * inRange(2n, 1n, 3n);
+ * ```
+ */
+function inRange(n, min, max) {
+    return isPosBig(n) && isPosBig(min) && isPosBig(max) && min <= n && n < max;
+}
+/**
+ * Asserts `min <= n < max`. NOTE: upper bound is exclusive.
+ * @param title - Value label for error messages.
+ * @param n - Candidate value.
+ * @param min - Inclusive lower bound.
+ * @param max - Exclusive upper bound.
+ * Wrong-type inputs are not separated from out-of-range values here: they still flow through the
+ * shared `RangeError` path because this is only a throwing wrapper around `inRange(...)`.
+ * @throws On wrong argument ranges or values. {@link RangeError}
+ * @example
+ * Assert that a bigint stays within one half-open range.
+ *
+ * ```ts
+ * aInRange('x', 2n, 1n, 256n);
+ * ```
+ */
+function aInRange(title, n, min, max) {
+    // Why min <= n < max and not a (min < n < max) OR b (min <= n <= max)?
+    // consider P=256n, min=0n, max=P
+    // - a for min=0 would require -1:          `inRange('x', x, -1n, P)`
+    // - b would commonly require subtraction:  `inRange('x', x, 0n, P - 1n)`
+    // - our way is the cleanest:               `inRange('x', x, 0n, P)
+    if (!inRange(n, min, max))
+        throw new RangeError('expected valid ' + title + ': ' + min + ' <= n < ' + max + ', got ' + n);
+}
+// Bit operations
+/**
+ * Calculates amount of bits in a bigint.
+ * Same as `n.toString(2).length`
+ * TODO: merge with nLength in modular
+ * @param n - Value to inspect.
+ * @returns Bit length.
+ * @throws If the value is negative. {@link Error}
+ * @example
+ * Measure the bit length of a scalar before serialization.
+ *
+ * ```ts
+ * bitLen(8n);
+ * ```
+ */
+function bitLen(n) {
+    // Size callers in this repo only use non-negative orders / scalars, so negative inputs are a
+    // contract bug and must not silently collapse to zero bits.
+    if (n < _0n$3)
+        throw new Error('expected non-negative bigint, got ' + n);
+    let len;
+    for (len = 0; n > _0n$3; n >>= _1n$3, len += 1)
+        ;
+    return len;
+}
+/**
+ * Calculate mask for N bits. Not using ** operator with bigints because of old engines.
+ * Same as BigInt(`0b${Array(i).fill('1').join('')}`)
+ * @param n - Number of bits. Negative widths are currently passed through to raw bigint shift
+ *   semantics and therefore produce `-1n`.
+ * @returns Bitmask value.
+ * @example
+ * Calculate mask for N bits.
+ *
+ * ```ts
+ * bitMask(4);
+ * ```
+ */
+const bitMask = (n) => (_1n$3 << BigInt(n)) - _1n$3;
+/**
+ * Minimal HMAC-DRBG from NIST 800-90 for RFC6979 sigs.
+ * @param hashLen - Hash output size in bytes. Callers are expected to pass a positive length; `0`
+ *   is not rejected here and would make the internal generate loop non-progressing.
+ * @param qByteLen - Requested output size in bytes. Callers are expected to pass a positive length.
+ * @param hmacFn - HMAC implementation.
+ * @returns Function that will call DRBG until the predicate returns anything
+ *   other than `undefined`.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Build a deterministic nonce generator for RFC6979-style signing.
+ *
+ * ```ts
+ * import { createHmacDrbg } from '@noble/curves/utils.js';
+ * import { hmac } from '@noble/hashes/hmac.js';
+ * import { sha256 } from '@noble/hashes/sha2.js';
+ * const drbg = createHmacDrbg(32, 32, (key, msg) => hmac(sha256, key, msg));
+ * const seed = new Uint8Array(32);
+ * drbg(seed, (bytes) => bytes);
+ * ```
+ */
+function createHmacDrbg(hashLen, qByteLen, hmacFn) {
+    anumber$1(hashLen, 'hashLen');
+    anumber$1(qByteLen, 'qByteLen');
+    if (typeof hmacFn !== 'function')
+        throw new TypeError('hmacFn must be a function');
+    // creates Uint8Array
+    const u8n = (len) => new Uint8Array(len);
+    const NULL = Uint8Array.of();
+    const byte0 = Uint8Array.of(0x00);
+    const byte1 = Uint8Array.of(0x01);
+    const _maxDrbgIters = 1000;
+    // Step B, Step C: set hashLen to 8*ceil(hlen/8).
+    // Minimal non-full-spec HMAC-DRBG from NIST 800-90 for RFC6979 signatures.
+    let v = u8n(hashLen);
+    // Steps B and C of RFC6979 3.2.
+    let k = u8n(hashLen);
+    let i = 0; // Iterations counter, will throw when over 1000
+    const reset = () => {
+        v.fill(1);
+        k.fill(0);
+        i = 0;
+    };
+    // hmac(k)(v, ...values)
+    const h = (...msgs) => hmacFn(k, concatBytes(v, ...msgs));
+    const reseed = (seed = NULL) => {
+        // HMAC-DRBG reseed() function. Steps D-G
+        k = h(byte0, seed); // k = hmac(k || v || 0x00 || seed)
+        v = h(); // v = hmac(k || v)
+        if (seed.length === 0)
+            return;
+        k = h(byte1, seed); // k = hmac(k || v || 0x01 || seed)
+        v = h(); // v = hmac(k || v)
+    };
+    const gen = () => {
+        // HMAC-DRBG generate() function
+        if (i++ >= _maxDrbgIters)
+            throw new Error('drbg: tried max amount of iterations');
+        let len = 0;
+        const out = [];
+        while (len < qByteLen) {
+            v = h();
+            const sl = v.slice();
+            out.push(sl);
+            len += v.length;
+        }
+        return concatBytes(...out);
+    };
+    const genUntil = (seed, pred) => {
+        reset();
+        reseed(seed); // Steps D-G
+        let res = undefined; // Step H: grind until the predicate accepts a candidate.
+        // Falsy values like 0 are valid outputs.
+        while ((res = pred(gen())) === undefined)
+            reseed();
+        reset();
+        return res;
+    };
+    return genUntil;
+}
+/**
+ * Validates declared required and optional field types on a plain object.
+ * Extra keys are intentionally ignored because many callers validate only the subset they use from
+ * richer option bags or runtime objects.
+ * @param object - Object to validate.
+ * @param fields - Required field types.
+ * @param optFields - Optional field types.
+ * @throws On wrong argument types. {@link TypeError}
+ * @example
+ * Check user options before building a curve helper.
+ *
+ * ```ts
+ * validateObject({ flag: true }, { flag: 'boolean' });
+ * ```
+ */
+function validateObject(object, fields = {}, optFields = {}) {
+    if (Object.prototype.toString.call(object) !== '[object Object]')
+        throw new TypeError('expected valid options object');
+    function checkField(fieldName, expectedType, isOpt) {
+        // Config/data fields must be explicit own properties, but runtime objects such as Field
+        // instances intentionally satisfy required method slots via their shared prototype.
+        if (!isOpt && expectedType !== 'function' && !Object.hasOwn(object, fieldName))
+            throw new TypeError(`param "${fieldName}" is invalid: expected own property`);
+        const val = object[fieldName];
+        if (isOpt && val === undefined)
+            return;
+        const current = typeof val;
+        if (current !== expectedType || val === null)
+            throw new TypeError(`param "${fieldName}" is invalid: expected ${expectedType}, got ${current}`);
+    }
+    const iter = (f, isOpt) => Object.entries(f).forEach(([k, v]) => checkField(k, v, isOpt));
+    iter(fields, false);
+    iter(optFields, true);
+}
+
+/**
+ * Utils for modular division and fields.
+ * Field over 11 is a finite (Galois) field is integer number operations `mod 11`.
+ * There is no division: it is replaced by modular multiplicative inverse.
+ * @module
+ */
+/*! noble-curves - MIT License (c) 2022 Paul Miller (paulmillr.com) */
+// Numbers aren't used in x25519 / x448 builds
+// prettier-ignore
+const _0n$2 = /* @__PURE__ */ BigInt(0), _1n$2 = /* @__PURE__ */ BigInt(1), _2n$2 = /* @__PURE__ */ BigInt(2);
+// prettier-ignore
+const _3n$1 = /* @__PURE__ */ BigInt(3), _4n$1 = /* @__PURE__ */ BigInt(4), _5n = /* @__PURE__ */ BigInt(5);
+// prettier-ignore
+const _7n = /* @__PURE__ */ BigInt(7), _8n = /* @__PURE__ */ BigInt(8), _9n = /* @__PURE__ */ BigInt(9);
+const _16n = /* @__PURE__ */ BigInt(16);
+/**
+ * @param a - Dividend value.
+ * @param b - Positive modulus.
+ * @returns Reduced value in `[0, b)` only when `b` is positive.
+ * @throws If the modulus is not positive. {@link Error}
+ * @example
+ * Normalize a bigint into one field residue.
+ *
+ * ```ts
+ * mod(-1n, 5n);
+ * ```
+ */
+function mod(a, b) {
+    if (b <= _0n$2)
+        throw new Error('mod: expected positive modulus, got ' + b);
+    const result = a % b;
+    return result >= _0n$2 ? result : b + result;
+}
+/**
+ * Does `x^(2^power)` mod p. `pow2(30, 4)` == `30^(2^4)`.
+ * Low-level helper: callers that need canonical residues must pass a valid `x` for the chosen
+ * modulus; the `power===0` fast path intentionally returns the input unchanged.
+ * @param x - Base value.
+ * @param power - Number of squarings.
+ * @param modulo - Reduction modulus.
+ * @returns Repeated-squaring result.
+ * @throws If the exponent is negative. {@link Error}
+ * @example
+ * Apply repeated squaring inside one field.
+ *
+ * ```ts
+ * pow2(3n, 2n, 11n);
+ * ```
+ */
+function pow2(x, power, modulo) {
+    if (power < _0n$2)
+        throw new Error('pow2: expected non-negative exponent, got ' + power);
+    let res = x;
+    while (power-- > _0n$2) {
+        res *= res;
+        res %= modulo;
+    }
+    return res;
+}
+/**
+ * Inverses number over modulo.
+ * Implemented using the {@link https://brilliant.org/wiki/extended-euclidean-algorithm/ | extended Euclidean algorithm}.
+ * @param number - Value to invert.
+ * @param modulo - Positive modulus.
+ * @returns Multiplicative inverse.
+ * @throws If the modulus is invalid or the inverse does not exist. {@link Error}
+ * @example
+ * Compute one modular inverse with the extended Euclidean algorithm.
+ *
+ * ```ts
+ * invert(3n, 11n);
+ * ```
+ */
+function invert(number, modulo) {
+    if (number === _0n$2)
+        throw new Error('invert: expected non-zero number');
+    if (modulo <= _0n$2)
+        throw new Error('invert: expected positive modulus, got ' + modulo);
+    // Fermat's little theorem "CT-like" version inv(n) = n^(m-2) mod m is 30x slower.
+    let a = mod(number, modulo);
+    let b = modulo;
+    // prettier-ignore
+    let x = _0n$2, u = _1n$2;
+    while (a !== _0n$2) {
+        const q = b / a;
+        const r = b - a * q;
+        const m = x - u * q;
+        // prettier-ignore
+        b = a, a = r, x = u, u = m;
+    }
+    const gcd = b;
+    if (gcd !== _1n$2)
+        throw new Error('invert: does not exist');
+    return mod(x, modulo);
+}
+function assertIsSquare(Fp, root, n) {
+    const F = Fp;
+    if (!F.eql(F.sqr(root), n))
+        throw new Error('Cannot find square root');
+}
+// Not all roots are possible! Example which will throw:
+// const NUM =
+// n = 72057594037927816n;
+// Fp = Field(BigInt('0x1a0111ea397fe69a4b1ba7b6434bacd764774b84f38512bf6730d2a0f6b0f6241eabfffeb153ffffb9feffffffffaaab'));
+function sqrt3mod4(Fp, n) {
+    const F = Fp;
+    const p1div4 = (F.ORDER + _1n$2) / _4n$1;
+    const root = F.pow(n, p1div4);
+    assertIsSquare(F, root, n);
+    return root;
+}
+// Equivalent `q = 5 (mod 8)` square-root formula (Atkin-style), not the RFC Appendix I.2 CMOV
+// pseudocode verbatim.
+function sqrt5mod8(Fp, n) {
+    const F = Fp;
+    const p5div8 = (F.ORDER - _5n) / _8n;
+    const n2 = F.mul(n, _2n$2);
+    const v = F.pow(n2, p5div8);
+    const nv = F.mul(n, v);
+    const i = F.mul(F.mul(nv, _2n$2), v);
+    const root = F.mul(nv, F.sub(i, F.ONE));
+    assertIsSquare(F, root, n);
+    return root;
+}
+// Based on RFC9380, Kong algorithm
+// prettier-ignore
+function sqrt9mod16(P) {
+    const Fp_ = Field(P);
+    const tn = tonelliShanks(P);
+    const c1 = tn(Fp_, Fp_.neg(Fp_.ONE)); //  1. c1 = sqrt(-1) in F, i.e., (c1^2) == -1 in F
+    const c2 = tn(Fp_, c1); //  2. c2 = sqrt(c1) in F, i.e., (c2^2) == c1 in F
+    const c3 = tn(Fp_, Fp_.neg(c1)); //  3. c3 = sqrt(-c1) in F, i.e., (c3^2) == -c1 in F
+    const c4 = (P + _7n) / _16n; //  4. c4 = (q + 7) / 16        # Integer arithmetic
+    return ((Fp, n) => {
+        const F = Fp;
+        let tv1 = F.pow(n, c4); //  1. tv1 = x^c4
+        let tv2 = F.mul(tv1, c1); //  2. tv2 = c1 * tv1
+        const tv3 = F.mul(tv1, c2); //  3. tv3 = c2 * tv1
+        const tv4 = F.mul(tv1, c3); //  4. tv4 = c3 * tv1
+        const e1 = F.eql(F.sqr(tv2), n); //  5.  e1 = (tv2^2) == x
+        const e2 = F.eql(F.sqr(tv3), n); //  6.  e2 = (tv3^2) == x
+        tv1 = F.cmov(tv1, tv2, e1); //  7. tv1 = CMOV(tv1, tv2, e1)  # Select tv2 if (tv2^2) == x
+        tv2 = F.cmov(tv4, tv3, e2); //  8. tv2 = CMOV(tv4, tv3, e2)  # Select tv3 if (tv3^2) == x
+        const e3 = F.eql(F.sqr(tv2), n); //  9.  e3 = (tv2^2) == x
+        const root = F.cmov(tv1, tv2, e3); // 10.  z = CMOV(tv1, tv2, e3)   # Select sqrt from tv1 & tv2
+        assertIsSquare(F, root, n);
+        return root;
+    });
+}
+/**
+ * Tonelli-Shanks square root search algorithm.
+ * This implementation is variable-time: it searches data-dependently for the first non-residue `Z`
+ * and for the smallest `i` in the main loop, unlike RFC 9380 Appendix I.4's constant-time shape.
+ * 1. {@link https://eprint.iacr.org/2012/685.pdf | eprint 2012/685}, page 12
+ * 2. Square Roots from 1; 24, 51, 10 to Dan Shanks
+ * @param P - field order
+ * @returns function that takes field Fp (created from P) and number n
+ * @throws If the field is too small, non-prime, or the square root does not exist. {@link Error}
+ * @example
+ * Construct a square-root helper for primes that need Tonelli-Shanks.
+ *
+ * ```ts
+ * import { Field, tonelliShanks } from '@noble/curves/abstract/modular.js';
+ * const Fp = Field(17n);
+ * const sqrt = tonelliShanks(17n)(Fp, 4n);
+ * ```
+ */
+function tonelliShanks(P) {
+    // Initialization (precomputation).
+    // Caching initialization could boost perf by 7%.
+    if (P < _3n$1)
+        throw new Error('sqrt is not defined for small field');
+    // Factor P - 1 = Q * 2^S, where Q is odd
+    let Q = P - _1n$2;
+    let S = 0;
+    while (Q % _2n$2 === _0n$2) {
+        Q /= _2n$2;
+        S++;
+    }
+    // Find the first quadratic non-residue Z >= 2
+    let Z = _2n$2;
+    const _Fp = Field(P);
+    while (FpLegendre(_Fp, Z) === 1) {
+        // Basic primality test for P. After x iterations, chance of
+        // not finding quadratic non-residue is 2^x, so 2^1000.
+        if (Z++ > 1000)
+            throw new Error('Cannot find square root: probably non-prime P');
+    }
+    // Fast-path; usually done before Z, but we do "primality test".
+    if (S === 1)
+        return sqrt3mod4;
+    // Slow-path
+    // TODO: test on Fp2 and others
+    let cc = _Fp.pow(Z, Q); // c = z^Q
+    const Q1div2 = (Q + _1n$2) / _2n$2;
+    return function tonelliSlow(Fp, n) {
+        const F = Fp;
+        if (F.is0(n))
+            return n;
+        // Check if n is a quadratic residue using Legendre symbol
+        if (FpLegendre(F, n) !== 1)
+            throw new Error('Cannot find square root');
+        // Initialize variables for the main loop
+        let M = S;
+        let c = F.mul(F.ONE, cc); // c = z^Q, move cc from field _Fp into field Fp
+        let t = F.pow(n, Q); // t = n^Q, first guess at the fudge factor
+        let R = F.pow(n, Q1div2); // R = n^((Q+1)/2), first guess at the square root
+        // Main loop
+        // while t != 1
+        while (!F.eql(t, F.ONE)) {
+            if (F.is0(t))
+                return F.ZERO; // if t=0 return R=0
+            let i = 1;
+            // Find the smallest i >= 1 such that t^(2^i) ≡ 1 (mod P)
+            let t_tmp = F.sqr(t); // t^(2^1)
+            while (!F.eql(t_tmp, F.ONE)) {
+                i++;
+                t_tmp = F.sqr(t_tmp); // t^(2^2)...
+                if (i === M)
+                    throw new Error('Cannot find square root');
+            }
+            // Calculate the exponent for b: 2^(M - i - 1)
+            const exponent = _1n$2 << BigInt(M - i - 1); // bigint is important
+            const b = F.pow(c, exponent); // b = 2^(M - i - 1)
+            // Update variables
+            M = i;
+            c = F.sqr(b); // c = b^2
+            t = F.mul(t, c); // t = (t * b^2)
+            R = F.mul(R, b); // R = R*b
+        }
+        return R;
+    };
+}
+/**
+ * Square root for a finite field. Will try optimized versions first:
+ *
+ * 1. P ≡ 3 (mod 4)
+ * 2. P ≡ 5 (mod 8)
+ * 3. P ≡ 9 (mod 16)
+ * 4. Tonelli-Shanks algorithm
+ *
+ * Different algorithms can give different roots, it is up to user to decide which one they want.
+ * For example there is FpSqrtOdd/FpSqrtEven to choose a root by oddness
+ * (used for hash-to-curve).
+ * @param P - Field order.
+ * @returns Square-root helper. The generic fallback inherits Tonelli-Shanks' variable-time
+ *   behavior and this selector assumes prime-field-style integer moduli.
+ * @throws If the field is unsupported or the square root does not exist. {@link Error}
+ * @example
+ * Choose the square-root helper appropriate for one field modulus.
+ *
+ * ```ts
+ * import { Field, FpSqrt } from '@noble/curves/abstract/modular.js';
+ * const Fp = Field(17n);
+ * const sqrt = FpSqrt(17n)(Fp, 4n);
+ * ```
+ */
+function FpSqrt(P) {
+    // P ≡ 3 (mod 4) => √n = n^((P+1)/4)
+    if (P % _4n$1 === _3n$1)
+        return sqrt3mod4;
+    // P ≡ 5 (mod 8) => Atkin algorithm, page 10 of https://eprint.iacr.org/2012/685.pdf
+    if (P % _8n === _5n)
+        return sqrt5mod8;
+    // P ≡ 9 (mod 16) => Kong algorithm, page 11 of https://eprint.iacr.org/2012/685.pdf (algorithm 4)
+    if (P % _16n === _9n)
+        return sqrt9mod16(P);
+    // Tonelli-Shanks algorithm
+    return tonelliShanks(P);
+}
+// prettier-ignore
+// Arithmetic-only subset checked by validateField(). This is intentionally not the full runtime
+// IField contract: helpers like `isValidNot0`, `invertBatch`, `toBytes`, `fromBytes`, `cmov`, and
+// field-specific extras like `isOdd` are left to the callers that actually need them.
+const FIELD_FIELDS = [
+    'create', 'isValid', 'is0', 'neg', 'inv', 'sqrt', 'sqr',
+    'eql', 'add', 'sub', 'mul', 'pow', 'div',
+    'addN', 'subN', 'mulN', 'sqrN'
+];
+/**
+ * @param field - Field implementation.
+ * @returns Validated field. This only checks the arithmetic subset needed by generic helpers; it
+ *   does not guarantee full runtime-method coverage for serialization, batching, `cmov`, or
+ *   field-specific extras beyond positive `BYTES` / `BITS`.
+ * @throws If the field shape or numeric metadata are invalid. {@link Error}
+ * @example
+ * Check that a field implementation exposes the operations curve code expects.
+ *
+ * ```ts
+ * import { Field, validateField } from '@noble/curves/abstract/modular.js';
+ * const Fp = validateField(Field(17n));
+ * ```
+ */
+function validateField(field) {
+    const initial = {
+        ORDER: 'bigint',
+        BYTES: 'number',
+        BITS: 'number',
+    };
+    const opts = FIELD_FIELDS.reduce((map, val) => {
+        map[val] = 'function';
+        return map;
+    }, initial);
+    validateObject(field, opts);
+    // Runtime field implementations must expose real integer byte/bit sizes; fractional / NaN /
+    // infinite metadata leaks through validateObject(type='number') but breaks encoders and caches.
+    asafenumber(field.BYTES, 'BYTES');
+    asafenumber(field.BITS, 'BITS');
+    // Runtime field implementations must expose positive byte/bit sizes; zero leaks through the
+    // numeric shape checks above but still breaks encoding helpers and cached-length assumptions.
+    if (field.BYTES < 1 || field.BITS < 1)
+        throw new Error('invalid field: expected BYTES/BITS > 0');
+    if (field.ORDER <= _1n$2)
+        throw new Error('invalid field: expected ORDER > 1, got ' + field.ORDER);
+    return field;
+}
+// Generic field functions
+/**
+ * Same as `pow` but for Fp: non-constant-time.
+ * Unsafe in some contexts: uses ladder, so can expose bigint bits.
+ * @param Fp - Field implementation.
+ * @param num - Base value.
+ * @param power - Exponent value.
+ * @returns Powered field element.
+ * @throws If the exponent is negative. {@link Error}
+ * @example
+ * Raise one field element to a public exponent.
+ *
+ * ```ts
+ * import { Field, FpPow } from '@noble/curves/abstract/modular.js';
+ * const Fp = Field(17n);
+ * const x = FpPow(Fp, 3n, 5n);
+ * ```
+ */
+function FpPow(Fp, num, power) {
+    const F = Fp;
+    if (power < _0n$2)
+        throw new Error('invalid exponent, negatives unsupported');
+    if (power === _0n$2)
+        return F.ONE;
+    if (power === _1n$2)
+        return num;
+    let p = F.ONE;
+    let d = num;
+    while (power > _0n$2) {
+        if (power & _1n$2)
+            p = F.mul(p, d);
+        d = F.sqr(d);
+        power >>= _1n$2;
+    }
+    return p;
+}
+/**
+ * Efficiently invert an array of Field elements.
+ * Exception-free. Zero-valued field elements stay `undefined` unless `passZero` is enabled.
+ * @param Fp - Field implementation.
+ * @param nums - Values to invert.
+ * @param passZero - map 0 to 0 (instead of undefined)
+ * @returns Inverted values.
+ * @example
+ * Invert several field elements with one shared inversion.
+ *
+ * ```ts
+ * import { Field, FpInvertBatch } from '@noble/curves/abstract/modular.js';
+ * const Fp = Field(17n);
+ * const inv = FpInvertBatch(Fp, [1n, 2n, 4n]);
+ * ```
+ */
+function FpInvertBatch(Fp, nums, passZero = false) {
+    const F = Fp;
+    const inverted = new Array(nums.length).fill(passZero ? F.ZERO : undefined);
+    // Walk from first to last, multiply them by each other MOD p
+    const multipliedAcc = nums.reduce((acc, num, i) => {
+        if (F.is0(num))
+            return acc;
+        inverted[i] = acc;
+        return F.mul(acc, num);
+    }, F.ONE);
+    // Invert last element
+    const invertedAcc = F.inv(multipliedAcc);
+    // Walk from last to first, multiply them by inverted each other MOD p
+    nums.reduceRight((acc, num, i) => {
+        if (F.is0(num))
+            return acc;
+        inverted[i] = F.mul(acc, inverted[i]);
+        return F.mul(acc, num);
+    }, invertedAcc);
+    return inverted;
+}
+/**
+ * Legendre symbol.
+ * Legendre constant is used to calculate Legendre symbol (a | p)
+ * which denotes the value of a^((p-1)/2) (mod p).
+ *
+ * * (a | p) ≡ 1    if a is a square (mod p), quadratic residue
+ * * (a | p) ≡ -1   if a is not a square (mod p), quadratic non residue
+ * * (a | p) ≡ 0    if a ≡ 0 (mod p)
+ * @param Fp - Field implementation.
+ * @param n - Value to inspect.
+ * @returns Legendre symbol.
+ * @throws If the field returns an invalid Legendre symbol value. {@link Error}
+ * @example
+ * Compute the Legendre symbol of one field element.
+ *
+ * ```ts
+ * import { Field, FpLegendre } from '@noble/curves/abstract/modular.js';
+ * const Fp = Field(17n);
+ * const symbol = FpLegendre(Fp, 4n);
+ * ```
+ */
+function FpLegendre(Fp, n) {
+    const F = Fp;
+    // We can use 3rd argument as optional cache of this value
+    // but seems unneeded for now. The operation is very fast.
+    const p1mod2 = (F.ORDER - _1n$2) / _2n$2;
+    const powered = F.pow(n, p1mod2);
+    const yes = F.eql(powered, F.ONE);
+    const zero = F.eql(powered, F.ZERO);
+    const no = F.eql(powered, F.neg(F.ONE));
+    if (!yes && !zero && !no)
+        throw new Error('invalid Legendre symbol result');
+    return yes ? 1 : zero ? 0 : -1;
+}
+/**
+ * @param n - Curve order. Callers are expected to pass a positive order.
+ * @param nBitLength - Optional cached bit length. Callers are expected to pass a positive cached
+ *   value when overriding the derived bit length.
+ * @returns Byte and bit lengths.
+ * @throws If the order or cached bit length is invalid. {@link Error}
+ * @example
+ * Measure the encoding sizes needed for one modulus.
+ *
+ * ```ts
+ * nLength(255n);
+ * ```
+ */
+function nLength(n, nBitLength) {
+    // Bit size, byte size of CURVE.n
+    if (nBitLength !== undefined)
+        anumber(nBitLength);
+    if (n <= _0n$2)
+        throw new Error('invalid n length: expected positive n, got ' + n);
+    if (nBitLength !== undefined && nBitLength < 1)
+        throw new Error('invalid n length: expected positive bit length, got ' + nBitLength);
+    const bits = bitLen(n);
+    // Cached bit lengths smaller than ORDER would truncate serialized scalars/elements and poison
+    // any math that relies on the derived field metadata.
+    if (nBitLength !== undefined && nBitLength < bits)
+        throw new Error(`invalid n length: expected bit length (${bits}) >= n.length (${nBitLength})`);
+    const _nBitLength = nBitLength !== undefined ? nBitLength : bits;
+    const nByteLength = Math.ceil(_nBitLength / 8);
+    return { nBitLength: _nBitLength, nByteLength };
+}
+// Keep the lazy sqrt cache off-instance so Field(...) can return a frozen object. Otherwise the
+// cached helper write would keep the field surface externally mutable.
+const FIELD_SQRT = new WeakMap();
+class _Field {
+    ORDER;
+    BITS;
+    BYTES;
+    isLE;
+    ZERO = _0n$2;
+    ONE = _1n$2;
+    _lengths;
+    _mod;
+    constructor(ORDER, opts = {}) {
+        // ORDER <= 1 is degenerate: ONE would not be a valid field element and helpers like pow/inv
+        // would stop modeling field arithmetic.
+        if (ORDER <= _1n$2)
+            throw new Error('invalid field: expected ORDER > 1, got ' + ORDER);
+        let _nbitLength = undefined;
+        this.isLE = false;
+        if (opts != null && typeof opts === 'object') {
+            // Cached bit lengths are trusted here and should already be positive / consistent with ORDER.
+            if (typeof opts.BITS === 'number')
+                _nbitLength = opts.BITS;
+            if (typeof opts.sqrt === 'function')
+                // `_Field.prototype` is frozen below, so custom sqrt hooks must become own properties
+                // explicitly instead of relying on writable prototype shadowing via assignment.
+                Object.defineProperty(this, 'sqrt', { value: opts.sqrt, enumerable: true });
+            if (typeof opts.isLE === 'boolean')
+                this.isLE = opts.isLE;
+            if (opts.allowedLengths)
+                this._lengths = Object.freeze(opts.allowedLengths.slice());
+            if (typeof opts.modFromBytes === 'boolean')
+                this._mod = opts.modFromBytes;
+        }
+        const { nBitLength, nByteLength } = nLength(ORDER, _nbitLength);
+        if (nByteLength > 2048)
+            throw new Error('invalid field: expected ORDER of <= 2048 bytes');
+        this.ORDER = ORDER;
+        this.BITS = nBitLength;
+        this.BYTES = nByteLength;
+        Object.freeze(this);
+    }
+    create(num) {
+        return mod(num, this.ORDER);
+    }
+    isValid(num) {
+        if (typeof num !== 'bigint')
+            throw new TypeError('invalid field element: expected bigint, got ' + typeof num);
+        return _0n$2 <= num && num < this.ORDER; // 0 is valid element, but it's not invertible
+    }
+    is0(num) {
+        return num === _0n$2;
+    }
+    // is valid and invertible
+    isValidNot0(num) {
+        return !this.is0(num) && this.isValid(num);
+    }
+    isOdd(num) {
+        return (num & _1n$2) === _1n$2;
+    }
+    neg(num) {
+        return mod(-num, this.ORDER);
+    }
+    eql(lhs, rhs) {
+        return lhs === rhs;
+    }
+    sqr(num) {
+        return mod(num * num, this.ORDER);
+    }
+    add(lhs, rhs) {
+        return mod(lhs + rhs, this.ORDER);
+    }
+    sub(lhs, rhs) {
+        return mod(lhs - rhs, this.ORDER);
+    }
+    mul(lhs, rhs) {
+        return mod(lhs * rhs, this.ORDER);
+    }
+    pow(num, power) {
+        return FpPow(this, num, power);
+    }
+    div(lhs, rhs) {
+        return mod(lhs * invert(rhs, this.ORDER), this.ORDER);
+    }
+    // Same as above, but doesn't normalize
+    sqrN(num) {
+        return num * num;
+    }
+    addN(lhs, rhs) {
+        return lhs + rhs;
+    }
+    subN(lhs, rhs) {
+        return lhs - rhs;
+    }
+    mulN(lhs, rhs) {
+        return lhs * rhs;
+    }
+    inv(num) {
+        return invert(num, this.ORDER);
+    }
+    sqrt(num) {
+        // Caching sqrt helpers speeds up sqrt9mod16 by 5x and Tonelli-Shanks by about 10% without keeping
+        // the field instance itself mutable.
+        let sqrt = FIELD_SQRT.get(this);
+        if (!sqrt)
+            FIELD_SQRT.set(this, (sqrt = FpSqrt(this.ORDER)));
+        return sqrt(this, num);
+    }
+    toBytes(num) {
+        // Serialize fixed-width limbs without re-validating the field range. Callers that need a
+        // canonical encoding must pass a valid element; some protocols intentionally serialize raw
+        // residues here and reduce or validate them elsewhere.
+        return this.isLE ? numberToBytesLE(num, this.BYTES) : numberToBytesBE(num, this.BYTES);
+    }
+    fromBytes(bytes, skipValidation = false) {
+        abytes(bytes);
+        const { _lengths: allowedLengths, BYTES, isLE, ORDER, _mod: modFromBytes } = this;
+        if (allowedLengths) {
+            // `allowedLengths` must list real positive byte lengths; otherwise empty input would get
+            // padded into zero and silently decode as a field element.
+            if (bytes.length < 1 || !allowedLengths.includes(bytes.length) || bytes.length > BYTES) {
+                throw new Error('Field.fromBytes: expected ' + allowedLengths + ' bytes, got ' + bytes.length);
+            }
+            const padded = new Uint8Array(BYTES);
+            // isLE add 0 to right, !isLE to the left.
+            padded.set(bytes, isLE ? 0 : padded.length - bytes.length);
+            bytes = padded;
+        }
+        if (bytes.length !== BYTES)
+            throw new Error('Field.fromBytes: expected ' + BYTES + ' bytes, got ' + bytes.length);
+        let scalar = isLE ? bytesToNumberLE(bytes) : bytesToNumberBE(bytes);
+        if (modFromBytes)
+            scalar = mod(scalar, ORDER);
+        if (!skipValidation)
+            if (!this.isValid(scalar))
+                throw new Error('invalid field element: outside of range 0..ORDER');
+        // Range validation is optional here because some protocols intentionally decode raw residues
+        // and reduce or validate them elsewhere.
+        return scalar;
+    }
+    // TODO: we don't need it here, move out to separate fn
+    invertBatch(lst) {
+        return FpInvertBatch(this, lst);
+    }
+    // We can't move this out because Fp6, Fp12 implement it
+    // and it's unclear what to return in there.
+    cmov(a, b, condition) {
+        // Field elements have `isValid(...)`; the CMOV branch bit is a direct runtime input, so reject
+        // non-boolean selectors here instead of letting JS truthiness silently change arithmetic.
+        abool(condition, 'condition');
+        return condition ? b : a;
+    }
+}
+// Freeze the shared method surface too; otherwise callers can still poison every Field instance by
+// monkey-patching `_Field.prototype` even if each instance is frozen.
+Object.freeze(_Field.prototype);
+/**
+ * Creates a finite field. Major performance optimizations:
+ * * 1. Denormalized operations like mulN instead of mul.
+ * * 2. Identical object shape: never add or remove keys.
+ * * 3. Frozen stable object shape; the lazy sqrt cache lives in a module-level `WeakMap`.
+ * Fragile: always run a benchmark on a change.
+ * Security note: operations and low-level serializers like `toBytes` don't check `isValid` for
+ * all elements for performance and protocol-flexibility reasons; callers are responsible for
+ * supplying valid elements when they need canonical field behavior.
+ * This is low-level code, please make sure you know what you're doing.
+ *
+ * Note about field properties:
+ * * CHARACTERISTIC p = prime number, number of elements in main subgroup.
+ * * ORDER q = similar to cofactor in curves, may be composite `q = p^m`.
+ *
+ * @param ORDER - field order, probably prime, or could be composite
+ * @param opts - Field options such as bit length or endianness. See {@link FieldOpts}.
+ * @returns Frozen field instance with a stable object shape. This wrapper forwards `opts` straight
+ *   into `_Field`, so it inherits `_Field`'s assumptions about cached sizes and `allowedLengths`.
+ * @example
+ * Construct one prime field with optional overrides.
+ *
+ * ```ts
+ * Field(11n);
+ * ```
+ */
+function Field(ORDER, opts = {}) {
+    return new _Field(ORDER, opts);
+}
+/**
+ * Returns total number of bytes consumed by the field element.
+ * For example, 32 bytes for usual 256-bit weierstrass curve.
+ * @param fieldOrder - number of field elements, usually CURVE.n. Callers are expected to pass an
+ *   order greater than 1.
+ * @returns byte length of field
+ * @throws If the field order is not a bigint. {@link Error}
+ * @example
+ * Read the fixed-width byte length of one field.
+ *
+ * ```ts
+ * getFieldBytesLength(255n);
+ * ```
+ */
+function getFieldBytesLength(fieldOrder) {
+    if (typeof fieldOrder !== 'bigint')
+        throw new Error('field order must be bigint');
+    // Valid field elements are in 0..ORDER-1, so ORDER <= 1 would make the encoded range degenerate.
+    if (fieldOrder <= _1n$2)
+        throw new Error('field order must be greater than 1');
+    // Valid field elements are < ORDER, so the maximal encoded element is ORDER - 1.
+    const bitLength = bitLen(fieldOrder - _1n$2);
+    return Math.ceil(bitLength / 8);
+}
+/**
+ * Returns minimal amount of bytes that can be safely reduced
+ * by field order.
+ * Should be 2^-128 for 128-bit curve such as P256.
+ * This is the reduction / modulo-bias lower bound; higher-level helpers may still impose a larger
+ * absolute floor for policy reasons.
+ * @param fieldOrder - number of field elements greater than 1, usually CURVE.n.
+ * @returns byte length of target hash
+ * @throws If the field order is invalid. {@link Error}
+ * @example
+ * Compute the minimum hash length needed for field reduction.
+ *
+ * ```ts
+ * getMinHashLength(255n);
+ * ```
+ */
+function getMinHashLength(fieldOrder) {
+    const length = getFieldBytesLength(fieldOrder);
+    return length + Math.ceil(length / 2);
+}
+/**
+ * "Constant-time" private key generation utility.
+ * Can take (n + n/2) or more bytes of uniform input e.g. from CSPRNG or KDF
+ * and convert them into private scalar, with the modulo bias being negligible.
+ * Needs at least 48 bytes of input for 32-byte private key. The implementation also keeps a hard
+ * 16-byte minimum even when `getMinHashLength(...)` is smaller, so toy-small inputs do not look
+ * accidentally acceptable for real scalar derivation.
+ * See {@link https://research.kudelskisecurity.com/2020/07/28/the-definitive-guide-to-modulo-bias-and-how-to-avoid-it/ | Kudelski's modulo-bias guide},
+ * {@link https://csrc.nist.gov/publications/detail/fips/186/5/final | FIPS 186-5 appendix A.2}, and
+ * {@link https://www.rfc-editor.org/rfc/rfc9380#section-5 | RFC 9380 section 5}. Unlike RFC 9380
+ * `hash_to_field`, this helper intentionally maps into the non-zero private-scalar range `1..n-1`.
+ * @param key - Uniform input bytes.
+ * @param fieldOrder - Size of subgroup.
+ * @param isLE - interpret hash bytes as LE num
+ * @returns valid private scalar
+ * @throws If the hash length or field order is invalid for scalar reduction. {@link Error}
+ * @example
+ * Map hash output into a private scalar range.
+ *
+ * ```ts
+ * mapHashToField(new Uint8Array(48).fill(1), 255n);
+ * ```
+ */
+function mapHashToField(key, fieldOrder, isLE = false) {
+    abytes(key);
+    const len = key.length;
+    const fieldLen = getFieldBytesLength(fieldOrder);
+    const minLen = Math.max(getMinHashLength(fieldOrder), 16);
+    // No toy-small inputs: the helper is for real scalar derivation, not tiny test curves. No huge
+    // inputs: easier to reason about JS timing / allocation behavior.
+    if (len < minLen || len > 1024)
+        throw new Error('expected ' + minLen + '-1024 bytes of input, got ' + len);
+    const num = isLE ? bytesToNumberLE(key) : bytesToNumberBE(key);
+    // `mod(x, 11)` can sometimes produce 0. `mod(x, 10) + 1` is the same, but no 0
+    const reduced = mod(num, fieldOrder - _1n$2) + _1n$2;
+    return isLE ? numberToBytesLE(reduced, fieldLen) : numberToBytesBE(reduced, fieldLen);
+}
+
+/**
+ * Methods for elliptic curve multiplication by scalars.
+ * Contains wNAF, pippenger.
+ * @module
+ */
+/*! noble-curves - MIT License (c) 2022 Paul Miller (paulmillr.com) */
+const _0n$1 = /* @__PURE__ */ BigInt(0);
+const _1n$1 = /* @__PURE__ */ BigInt(1);
+/**
+ * Computes both candidates first, but the final selection still branches on `condition`, so this
+ * is not a strict constant-time CMOV primitive.
+ * @param condition - Whether to negate the point.
+ * @param item - Point-like value.
+ * @returns Original or negated value.
+ * @example
+ * Keep the point or return its negation based on one boolean branch.
+ *
+ * ```ts
+ * import { negateCt } from '@noble/curves/abstract/curve.js';
+ * import { p256 } from '@noble/curves/nist.js';
+ * const maybeNegated = negateCt(true, p256.Point.BASE);
+ * ```
+ */
+function negateCt(condition, item) {
+    const neg = item.negate();
+    return condition ? neg : item;
+}
+/**
+ * Takes a bunch of Projective Points but executes only one
+ * inversion on all of them. Inversion is very slow operation,
+ * so this improves performance massively.
+ * Optimization: converts a list of projective points to a list of identical points with Z=1.
+ * Input points are left unchanged; the normalized points are returned as fresh instances.
+ * @param c - Point constructor.
+ * @param points - Projective points.
+ * @returns Fresh projective points reconstructed from normalized affine coordinates.
+ * @example
+ * Batch-normalize projective points with a single shared inversion.
+ *
+ * ```ts
+ * import { normalizeZ } from '@noble/curves/abstract/curve.js';
+ * import { p256 } from '@noble/curves/nist.js';
+ * const points = normalizeZ(p256.Point, [p256.Point.BASE, p256.Point.BASE.double()]);
+ * ```
+ */
+function normalizeZ(c, points) {
+    const invertedZs = FpInvertBatch(c.Fp, points.map((p) => p.Z));
+    return points.map((p, i) => c.fromAffine(p.toAffine(invertedZs[i])));
+}
+function validateW(W, bits) {
+    if (!Number.isSafeInteger(W) || W <= 0 || W > bits)
+        throw new Error('invalid window size, expected [1..' + bits + '], got W=' + W);
+}
+function calcWOpts(W, scalarBits) {
+    validateW(W, scalarBits);
+    const windows = Math.ceil(scalarBits / W) + 1; // W=8 33. Not 32, because we skip zero
+    const windowSize = 2 ** (W - 1); // W=8 128. Not 256, because we skip zero
+    const maxNumber = 2 ** W; // W=8 256
+    const mask = bitMask(W); // W=8 255 == mask 0b11111111
+    const shiftBy = BigInt(W); // W=8 8
+    return { windows, windowSize, mask, maxNumber, shiftBy };
+}
+function calcOffsets(n, window, wOpts) {
+    const { windowSize, mask, maxNumber, shiftBy } = wOpts;
+    let wbits = Number(n & mask); // extract W bits.
+    let nextN = n >> shiftBy; // shift number by W bits.
+    // What actually happens here:
+    // const highestBit = Number(mask ^ (mask >> 1n));
+    // let wbits2 = wbits - 1; // skip zero
+    // if (wbits2 & highestBit) { wbits2 ^= Number(mask); // (~);
+    // split if bits > max: +224 => 256-32
+    if (wbits > windowSize) {
+        // we skip zero, which means instead of `>= size-1`, we do `> size`
+        wbits -= maxNumber; // -32, can be maxNumber - wbits, but then we need to set isNeg here.
+        nextN += _1n$1; // +256 (carry)
+    }
+    const offsetStart = window * windowSize;
+    const offset = offsetStart + Math.abs(wbits) - 1; // -1 because we skip zero; ignore when isZero
+    const isZero = wbits === 0; // is current window slice a 0?
+    const isNeg = wbits < 0; // is current window slice negative?
+    const isNegF = window % 2 !== 0; // fake branch noise only
+    const offsetF = offsetStart; // fake branch noise only
+    return { nextN, offset, isZero, isNeg, isNegF, offsetF };
+}
+// Since points in different groups cannot be equal (different object constructor),
+// we can have single place to store precomputes.
+// Allows to make points frozen / immutable.
+const pointPrecomputes = new WeakMap();
+const pointWindowSizes = new WeakMap();
+function getW(P) {
+    // To disable precomputes:
+    // return 1;
+    // `1` is also the uncached sentinel: use the ladder / non-precomputed path.
+    return pointWindowSizes.get(P) || 1;
+}
+function assert0(n) {
+    // Internal invariant: a non-zero remainder here means the wNAF window decomposition or loop
+    // count is inconsistent, not that the original caller provided a bad scalar.
+    if (n !== _0n$1)
+        throw new Error('invalid wNAF');
+}
+/**
+ * Elliptic curve multiplication of Point by scalar. Fragile.
+ * Table generation takes **30MB of ram and 10ms on high-end CPU**,
+ * but may take much longer on slow devices. Actual generation will happen on
+ * first call of `multiply()`. By default, `BASE` point is precomputed.
+ *
+ * Scalars should always be less than curve order: this should be checked inside of a curve itself.
+ * Creates precomputation tables for fast multiplication:
+ * - private scalar is split by fixed size windows of W bits
+ * - every window point is collected from window's table & added to accumulator
+ * - since windows are different, same point inside tables won't be accessed more than once per calc
+ * - each multiplication is 'Math.ceil(CURVE_ORDER / 𝑊) + 1' point additions (fixed for any scalar)
+ * - +1 window is neccessary for wNAF
+ * - wNAF reduces table size: 2x less memory + 2x faster generation, but 10% slower multiplication
+ *
+ * TODO: research returning a 2d JS array of windows instead of a single window.
+ * This would allow windows to be in different memory locations.
+ * @param Point - Point constructor.
+ * @param bits - Scalar bit length.
+ * @example
+ * Elliptic curve multiplication of Point by scalar.
+ *
+ * ```ts
+ * import { wNAF } from '@noble/curves/abstract/curve.js';
+ * import { p256 } from '@noble/curves/nist.js';
+ * const ladder = new wNAF(p256.Point, p256.Point.Fn.BITS);
+ * ```
+ */
+class wNAF {
+    BASE;
+    ZERO;
+    Fn;
+    bits;
+    // Parametrized with a given Point class (not individual point)
+    constructor(Point, bits) {
+        this.BASE = Point.BASE;
+        this.ZERO = Point.ZERO;
+        this.Fn = Point.Fn;
+        this.bits = bits;
+    }
+    // non-const time multiplication ladder
+    _unsafeLadder(elm, n, p = this.ZERO) {
+        let d = elm;
+        while (n > _0n$1) {
+            if (n & _1n$1)
+                p = p.add(d);
+            d = d.double();
+            n >>= _1n$1;
+        }
+        return p;
+    }
+    /**
+     * Creates a wNAF precomputation window. Used for caching.
+     * Default window size is set by `utils.precompute()` and is equal to 8.
+     * Number of precomputed points depends on the curve size:
+     * 2^(𝑊−1) * (Math.ceil(𝑛 / 𝑊) + 1), where:
+     * - 𝑊 is the window size
+     * - 𝑛 is the bitlength of the curve order.
+     * For a 256-bit curve and window size 8, the number of precomputed points is 128 * 33 = 4224.
+     * @param point - Point instance
+     * @param W - window size
+     * @returns precomputed point tables flattened to a single array
+     */
+    precomputeWindow(point, W) {
+        const { windows, windowSize } = calcWOpts(W, this.bits);
+        const points = [];
+        let p = point;
+        let base = p;
+        for (let window = 0; window < windows; window++) {
+            base = p;
+            points.push(base);
+            // i=1, bc we skip 0
+            for (let i = 1; i < windowSize; i++) {
+                base = base.add(p);
+                points.push(base);
+            }
+            p = base.double();
+        }
+        return points;
+    }
+    /**
+     * Implements ec multiplication using precomputed tables and w-ary non-adjacent form.
+     * More compact implementation:
+     * https://github.com/paulmillr/noble-secp256k1/blob/47cb1669b6e506ad66b35fe7d76132ae97465da2/index.ts#L502-L541
+     * @returns real and fake (for const-time) points
+     */
+    wNAF(W, precomputes, n) {
+        // Scalar should be smaller than field order
+        if (!this.Fn.isValid(n))
+            throw new Error('invalid scalar');
+        // Accumulators
+        let p = this.ZERO;
+        let f = this.BASE;
+        // This code was first written with assumption that 'f' and 'p' will never be infinity point:
+        // since each addition is multiplied by 2 ** W, it cannot cancel each other. However,
+        // there is negate now: it is possible that negated element from low value
+        // would be the same as high element, which will create carry into next window.
+        // It's not obvious how this can fail, but still worth investigating later.
+        const wo = calcWOpts(W, this.bits);
+        for (let window = 0; window < wo.windows; window++) {
+            // (n === _0n) is handled and not early-exited. isEven and offsetF are used for noise
+            const { nextN, offset, isZero, isNeg, isNegF, offsetF } = calcOffsets(n, window, wo);
+            n = nextN;
+            if (isZero) {
+                // bits are 0: add garbage to fake point
+                // Important part for const-time getPublicKey: add random "noise" point to f.
+                f = f.add(negateCt(isNegF, precomputes[offsetF]));
+            }
+            else {
+                // bits are 1: add to result point
+                p = p.add(negateCt(isNeg, precomputes[offset]));
+            }
+        }
+        assert0(n);
+        // Return both real and fake points so JIT keeps the noise path alive.
+        // Known caveat: negate/carry interactions can still drive `f` to infinity even when `p` is not,
+        // which weakens the noise path and leaves this only "less const-time" by about one bigint mul.
+        return { p, f };
+    }
+    /**
+     * Implements unsafe EC multiplication using precomputed tables
+     * and w-ary non-adjacent form.
+     * @param acc - accumulator point to add result of multiplication
+     * @returns point
+     */
+    wNAFUnsafe(W, precomputes, n, acc = this.ZERO) {
+        const wo = calcWOpts(W, this.bits);
+        for (let window = 0; window < wo.windows; window++) {
+            if (n === _0n$1)
+                break; // Early-exit, skip 0 value
+            const { nextN, offset, isZero, isNeg } = calcOffsets(n, window, wo);
+            n = nextN;
+            if (isZero) {
+                // Window bits are 0: skip processing.
+                // Move to next window.
+                continue;
+            }
+            else {
+                const item = precomputes[offset];
+                acc = acc.add(isNeg ? item.negate() : item); // Re-using acc allows to save adds in MSM
+            }
+        }
+        assert0(n);
+        return acc;
+    }
+    getPrecomputes(W, point, transform) {
+        // Cache key is only point identity plus the remembered window size; callers must not reuse the
+        // same point with incompatible `transform(...)` layouts and expect a separate cache entry.
+        let comp = pointPrecomputes.get(point);
+        if (!comp) {
+            comp = this.precomputeWindow(point, W);
+            if (W !== 1) {
+                // Doing transform outside of if brings 15% perf hit
+                if (typeof transform === 'function')
+                    comp = transform(comp);
+                pointPrecomputes.set(point, comp);
+            }
+        }
+        return comp;
+    }
+    cached(point, scalar, transform) {
+        const W = getW(point);
+        return this.wNAF(W, this.getPrecomputes(W, point, transform), scalar);
+    }
+    unsafe(point, scalar, transform, prev) {
+        const W = getW(point);
+        if (W === 1)
+            return this._unsafeLadder(point, scalar, prev); // For W=1 ladder is ~x2 faster
+        return this.wNAFUnsafe(W, this.getPrecomputes(W, point, transform), scalar, prev);
+    }
+    // We calculate precomputes for elliptic curve point multiplication
+    // using windowed method. This specifies window size and
+    // stores precomputed values. Usually only base point would be precomputed.
+    createCache(P, W) {
+        validateW(W, this.bits);
+        pointWindowSizes.set(P, W);
+        pointPrecomputes.delete(P);
+    }
+    hasCache(elm) {
+        return getW(elm) !== 1;
+    }
+}
+/**
+ * Endomorphism-specific multiplication for Koblitz curves.
+ * Cost: 128 dbl, 0-256 adds.
+ * @param Point - Point constructor.
+ * @param point - Input point.
+ * @param k1 - First non-negative absolute scalar chunk.
+ * @param k2 - Second non-negative absolute scalar chunk.
+ * @returns Partial multiplication results.
+ * @example
+ * Endomorphism-specific multiplication for Koblitz curves.
+ *
+ * ```ts
+ * import { mulEndoUnsafe } from '@noble/curves/abstract/curve.js';
+ * import { secp256k1 } from '@noble/curves/secp256k1.js';
+ * const parts = mulEndoUnsafe(secp256k1.Point, secp256k1.Point.BASE, 3n, 5n);
+ * ```
+ */
+function mulEndoUnsafe(Point, point, k1, k2) {
+    let acc = point;
+    let p1 = Point.ZERO;
+    let p2 = Point.ZERO;
+    while (k1 > _0n$1 || k2 > _0n$1) {
+        if (k1 & _1n$1)
+            p1 = p1.add(acc);
+        if (k2 & _1n$1)
+            p2 = p2.add(acc);
+        acc = acc.double();
+        k1 >>= _1n$1;
+        k2 >>= _1n$1;
+    }
+    return { p1, p2 };
+}
+function createField(order, field, isLE) {
+    if (field) {
+        // Reuse supplied field overrides as-is; `isLE` only affects freshly constructed fallback
+        // fields, and validateField() below only checks the arithmetic subset, not full byte/cmov
+        // behavior.
+        if (field.ORDER !== order)
+            throw new Error('Field.ORDER must match order: Fp == p, Fn == n');
+        validateField(field);
+        return field;
+    }
+    else {
+        return Field(order, { isLE });
+    }
+}
+/**
+ * Validates basic CURVE shape and field membership, then creates fields.
+ * This does not prove that the generator is on-curve, that subgroup/order data are consistent, or
+ * that the curve equation itself is otherwise sane.
+ * @param type - Curve family.
+ * @param CURVE - Curve parameters.
+ * @param curveOpts - Optional field overrides:
+ *   - `Fp` (optional): Optional base-field override.
+ *   - `Fn` (optional): Optional scalar-field override.
+ * @param FpFnLE - Whether field encoding is little-endian.
+ * @returns Frozen curve parameters and fields.
+ * @throws If the curve parameters or field overrides are invalid. {@link Error}
+ * @example
+ * Build curve fields from raw constants before constructing a curve instance.
+ *
+ * ```ts
+ * const curve = createCurveFields('weierstrass', {
+ *   p: 17n,
+ *   n: 19n,
+ *   h: 1n,
+ *   a: 2n,
+ *   b: 2n,
+ *   Gx: 5n,
+ *   Gy: 1n,
+ * });
+ * ```
+ */
+function createCurveFields(type, CURVE, curveOpts = {}, FpFnLE) {
+    if (FpFnLE === undefined)
+        FpFnLE = type === 'edwards';
+    if (!CURVE || typeof CURVE !== 'object')
+        throw new Error(`expected valid ${type} CURVE object`);
+    for (const p of ['p', 'n', 'h']) {
+        const val = CURVE[p];
+        if (!(typeof val === 'bigint' && val > _0n$1))
+            throw new Error(`CURVE.${p} must be positive bigint`);
+    }
+    const Fp = createField(CURVE.p, curveOpts.Fp, FpFnLE);
+    const Fn = createField(CURVE.n, curveOpts.Fn, FpFnLE);
+    const _b = 'b' ;
+    const params = ['Gx', 'Gy', 'a', _b];
+    for (const p of params) {
+        // @ts-ignore
+        if (!Fp.isValid(CURVE[p]))
+            throw new Error(`CURVE.${p} must be valid field element of CURVE.Fp`);
+    }
+    CURVE = Object.freeze(Object.assign({}, CURVE));
+    return { CURVE, Fp, Fn };
+}
+/**
+ * @param randomSecretKey - Secret-key generator.
+ * @param getPublicKey - Public-key derivation helper.
+ * @returns Keypair generator.
+ * @example
+ * Build a `keygen()` helper from existing secret-key and public-key primitives.
+ *
+ * ```ts
+ * import { createKeygen } from '@noble/curves/abstract/curve.js';
+ * import { p256 } from '@noble/curves/nist.js';
+ * const keygen = createKeygen(p256.utils.randomSecretKey, p256.getPublicKey);
+ * const pair = keygen();
+ * ```
+ */
+function createKeygen(randomSecretKey, getPublicKey) {
+    return function keygen(seed) {
+        const secretKey = randomSecretKey(seed);
+        return { secretKey, publicKey: getPublicKey(secretKey) };
+    };
+}
+
+/**
+ * HMAC: RFC2104 message authentication code.
+ * @module
+ */
+/**
+ * Internal class for HMAC.
+ * Accepts any byte key, although RFC 2104 §3 recommends keys at least
+ * `HashLen` bytes long.
+ */
+class _HMAC {
+    oHash;
+    iHash;
+    blockLen;
+    outputLen;
+    canXOF = false;
+    finished = false;
+    destroyed = false;
+    constructor(hash, key) {
+        ahash(hash);
+        abytes$1(key, undefined, 'key');
+        this.iHash = hash.create();
+        if (typeof this.iHash.update !== 'function')
+            throw new Error('Expected instance of class which extends utils.Hash');
+        this.blockLen = this.iHash.blockLen;
+        this.outputLen = this.iHash.outputLen;
+        const blockLen = this.blockLen;
+        const pad = new Uint8Array(blockLen);
+        // blockLen can be bigger than outputLen
+        pad.set(key.length > blockLen ? hash.create().update(key).digest() : key);
+        for (let i = 0; i < pad.length; i++)
+            pad[i] ^= 0x36;
+        this.iHash.update(pad);
+        // By doing update (processing of the first block) of the outer hash here,
+        // we can re-use it between multiple calls via clone.
+        this.oHash = hash.create();
+        // Undo internal XOR && apply outer XOR
+        for (let i = 0; i < pad.length; i++)
+            pad[i] ^= 0x36 ^ 0x5c;
+        this.oHash.update(pad);
+        clean(pad);
+    }
+    update(buf) {
+        aexists(this);
+        this.iHash.update(buf);
+        return this;
+    }
+    digestInto(out) {
+        aexists(this);
+        aoutput(out, this);
+        this.finished = true;
+        const buf = out.subarray(0, this.outputLen);
+        // Reuse the first outputLen bytes for the inner digest; the outer hash consumes them before
+        // overwriting that same prefix with the final tag, leaving any oversized tail untouched.
+        this.iHash.digestInto(buf);
+        this.oHash.update(buf);
+        this.oHash.digestInto(buf);
+        this.destroy();
+    }
+    digest() {
+        const out = new Uint8Array(this.oHash.outputLen);
+        this.digestInto(out);
+        return out;
+    }
+    _cloneInto(to) {
+        // Create new instance without calling constructor since the key
+        // is already in state and we don't know it.
+        to ||= Object.create(Object.getPrototypeOf(this), {});
+        const { oHash, iHash, finished, destroyed, blockLen, outputLen } = this;
+        to = to;
+        to.finished = finished;
+        to.destroyed = destroyed;
+        to.blockLen = blockLen;
+        to.outputLen = outputLen;
+        to.oHash = oHash._cloneInto(to.oHash);
+        to.iHash = iHash._cloneInto(to.iHash);
+        return to;
+    }
+    clone() {
+        return this._cloneInto();
+    }
+    destroy() {
+        this.destroyed = true;
+        this.oHash.destroy();
+        this.iHash.destroy();
+    }
+}
+const hmac = /* @__PURE__ */ (() => {
+    const hmac_ = ((hash, key, message) => new _HMAC(hash, key).update(message).digest());
+    hmac_.create = (hash, key) => new _HMAC(hash, key);
+    return hmac_;
+})();
+
+/**
+ * Short Weierstrass curve methods. The formula is: y² = x³ + ax + b.
+ *
+ * ### Design rationale for types
+ *
+ * * Interaction between classes from different curves should fail:
+ *   `k256.Point.BASE.add(p256.Point.BASE)`
+ * * For this purpose we want to use `instanceof` operator, which is fast and works during runtime
+ * * Different calls of `curve()` would return different classes -
+ *   `curve(params) !== curve(params)`: if somebody decided to monkey-patch their curve,
+ *   it won't affect others
+ *
+ * TypeScript can't infer types for classes created inside a function. Classes is one instance
+ * of nominative types in TypeScript and interfaces only check for shape, so it's hard to create
+ * unique type for every function call.
+ *
+ * We can use generic types via some param, like curve opts, but that would:
+ *     1. Enable interaction between `curve(params)` and `curve(params)` (curves of same params)
+ *     which is hard to debug.
+ *     2. Params can be generic and we can't enforce them to be constant value:
+ *     if somebody creates curve from non-constant params,
+ *     it would be allowed to interact with other curves with non-constant params
+ *
+ * @todo https://www.typescriptlang.org/docs/handbook/release-notes/typescript-2-7.html#unique-symbol
+ * @module
+ */
+/*! noble-curves - MIT License (c) 2022 Paul Miller (paulmillr.com) */
+// We construct the basis so `den` is always positive and equals `n`,
+// but the `num` sign depends on the basis, not on the secret value.
+// Exact half-way cases round away from zero, which keeps the split symmetric
+// around the reduced-basis boundaries used by endomorphism decomposition.
+const divNearest = (num, den) => (num + (num >= 0 ? den : -den) / _2n$1) / den;
+/** Splits scalar for GLV endomorphism. */
+function _splitEndoScalar(k, basis, n) {
+    // Split scalar into two such that part is ~half bits: `abs(part) < sqrt(N)`
+    // Since part can be negative, we need to do this on point.
+    // Callers must provide a reduced GLV basis whose vectors satisfy
+    // `a + b * lambda ≡ 0 (mod n)`; this helper only sees the basis and `n`.
+    // Reject unreduced scalars instead of silently treating them mod n.
+    aInRange('scalar', k, _0n, n);
+    // TODO: verifyScalar function which consumes lambda
+    const [[a1, b1], [a2, b2]] = basis;
+    const c1 = divNearest(b2 * k, n);
+    const c2 = divNearest(-b1 * k, n);
+    // |k1|/|k2| is < sqrt(N), but can be negative.
+    // If we do `k1 mod N`, we'll get big scalar (`> sqrt(N)`): so, we do cheaper negation instead.
+    let k1 = k - c1 * a1 - c2 * a2;
+    let k2 = -c1 * b1 - c2 * b2;
+    const k1neg = k1 < _0n;
+    const k2neg = k2 < _0n;
+    if (k1neg)
+        k1 = -k1;
+    if (k2neg)
+        k2 = -k2;
+    // Double check that resulting scalar less than half bits of N: otherwise wNAF will fail.
+    // This should only happen on wrong bases.
+    // Also, the math inside is complex enough that this guard is worth keeping.
+    const MAX_NUM = bitMask(Math.ceil(bitLen(n) / 2)) + _1n; // Half bits of N
+    if (k1 < _0n || k1 >= MAX_NUM || k2 < _0n || k2 >= MAX_NUM) {
+        throw new Error('splitScalar (endomorphism): failed for k');
+    }
+    return { k1neg, k1, k2neg, k2 };
+}
+function validateSigFormat(format) {
+    if (!['compact', 'recovered', 'der'].includes(format))
+        throw new Error('Signature format must be "compact", "recovered", or "der"');
+    return format;
+}
+function validateSigOpts(opts, def) {
+    validateObject(opts);
+    const optsn = {};
+    // Normalize only the declared option subset from `def`; unknown keys are
+    // intentionally ignored so shared / superset option bags stay valid here too.
+    // `extraEntropy` stays an opaque payload until the signing path consumes it.
+    for (let optName of Object.keys(def)) {
+        // @ts-ignore
+        optsn[optName] = opts[optName] === undefined ? def[optName] : opts[optName];
+    }
+    abool(optsn.lowS, 'lowS');
+    abool(optsn.prehash, 'prehash');
+    if (optsn.format !== undefined)
+        validateSigFormat(optsn.format);
+    return optsn;
+}
+/**
+ * @param m - Error message.
+ * @example
+ * Throw a DER-specific error when signature parsing encounters invalid bytes.
+ *
+ * ```ts
+ * new DERErr('bad der');
+ * ```
+ */
+class DERErr extends Error {
+    constructor(m = '') {
+        super(m);
+    }
+}
+/**
+ * ASN.1 DER encoding utilities. ASN is very complex & fragile. Format:
+ *
+ *     [0x30 (SEQUENCE), bytelength, 0x02 (INTEGER), intLength, R, 0x02 (INTEGER), intLength, S]
+ *
+ * Docs: {@link https://letsencrypt.org/docs/a-warm-welcome-to-asn1-and-der/ | Let's Encrypt ASN.1 guide} and
+ * {@link https://luca.ntop.org/Teaching/Appunti/asn1.html | Luca Deri's ASN.1 notes}.
+ * @example
+ * ASN.1 DER encoding utilities.
+ *
+ * ```ts
+ * const der = DER.hexFromSig({ r: 1n, s: 2n });
+ * ```
+ */
+const DER = {
+    // asn.1 DER encoding utils
+    Err: DERErr,
+    // Basic building block is TLV (Tag-Length-Value)
+    _tlv: {
+        encode: (tag, data) => {
+            const { Err: E } = DER;
+            asafenumber(tag, 'tag');
+            if (tag < 0 || tag > 255)
+                throw new E('tlv.encode: wrong tag');
+            if (typeof data !== 'string')
+                throw new TypeError('"data" expected string, got type=' + typeof data);
+            // Internal helper: callers hand this already-validated hex payload, so we only enforce
+            // byte alignment here instead of re-validating every nibble.
+            if (data.length & 1)
+                throw new E('tlv.encode: unpadded data');
+            const dataLen = data.length / 2;
+            const len = numberToHexUnpadded(dataLen);
+            if ((len.length / 2) & 0b1000_0000)
+                throw new E('tlv.encode: long form length too big');
+            // length of length with long form flag
+            const lenLen = dataLen > 127 ? numberToHexUnpadded((len.length / 2) | 0b1000_0000) : '';
+            const t = numberToHexUnpadded(tag);
+            return t + lenLen + len + data;
+        },
+        // v - value, l - left bytes (unparsed)
+        decode(tag, data) {
+            const { Err: E } = DER;
+            data = abytes(data, undefined, 'DER data');
+            let pos = 0;
+            if (tag < 0 || tag > 255)
+                throw new E('tlv.encode: wrong tag');
+            if (data.length < 2 || data[pos++] !== tag)
+                throw new E('tlv.decode: wrong tlv');
+            const first = data[pos++];
+            // First bit of first length byte is the short/long form flag.
+            const isLong = !!(first & 0b1000_0000);
+            let length = 0;
+            if (!isLong)
+                length = first;
+            else {
+                // Long form: [longFlag(1bit), lengthLength(7bit), length (BE)]
+                const lenLen = first & 0b0111_1111;
+                if (!lenLen)
+                    throw new E('tlv.decode(long): indefinite length not supported');
+                // This would overflow u32 in JS.
+                if (lenLen > 4)
+                    throw new E('tlv.decode(long): byte length is too big');
+                const lengthBytes = data.subarray(pos, pos + lenLen);
+                if (lengthBytes.length !== lenLen)
+                    throw new E('tlv.decode: length bytes not complete');
+                if (lengthBytes[0] === 0)
+                    throw new E('tlv.decode(long): zero leftmost byte');
+                for (const b of lengthBytes)
+                    length = (length << 8) | b;
+                pos += lenLen;
+                if (length < 128)
+                    throw new E('tlv.decode(long): not minimal encoding');
+            }
+            const v = data.subarray(pos, pos + length);
+            if (v.length !== length)
+                throw new E('tlv.decode: wrong value length');
+            return { v, l: data.subarray(pos + length) };
+        },
+    },
+    // https://crypto.stackexchange.com/a/57734 Leftmost bit of first byte is 'negative' flag,
+    // since we always use positive integers here. It must always be empty:
+    // - add zero byte if exists
+    // - if next byte doesn't have a flag, leading zero is not allowed (minimal encoding)
+    _int: {
+        encode(num) {
+            const { Err: E } = DER;
+            abignumber(num);
+            if (num < _0n)
+                throw new E('integer: negative integers are not allowed');
+            let hex = numberToHexUnpadded(num);
+            // Pad with zero byte if negative flag is present
+            if (Number.parseInt(hex[0], 16) & 0b1000)
+                hex = '00' + hex;
+            if (hex.length & 1)
+                throw new E('unexpected DER parsing assertion: unpadded hex');
+            return hex;
+        },
+        decode(data) {
+            const { Err: E } = DER;
+            if (data.length < 1)
+                throw new E('invalid signature integer: empty');
+            if (data[0] & 0b1000_0000)
+                throw new E('invalid signature integer: negative');
+            // Single-byte zero `00` is the canonical DER INTEGER encoding for zero.
+            if (data.length > 1 && data[0] === 0x00 && !(data[1] & 0b1000_0000))
+                throw new E('invalid signature integer: unnecessary leading zero');
+            return bytesToNumberBE(data);
+        },
+    },
+    toSig(bytes) {
+        // parse DER signature
+        const { Err: E, _int: int, _tlv: tlv } = DER;
+        const data = abytes(bytes, undefined, 'signature');
+        const { v: seqBytes, l: seqLeftBytes } = tlv.decode(0x30, data);
+        if (seqLeftBytes.length)
+            throw new E('invalid signature: left bytes after parsing');
+        const { v: rBytes, l: rLeftBytes } = tlv.decode(0x02, seqBytes);
+        const { v: sBytes, l: sLeftBytes } = tlv.decode(0x02, rLeftBytes);
+        if (sLeftBytes.length)
+            throw new E('invalid signature: left bytes after parsing');
+        return { r: int.decode(rBytes), s: int.decode(sBytes) };
+    },
+    hexFromSig(sig) {
+        const { _tlv: tlv, _int: int } = DER;
+        const rs = tlv.encode(0x02, int.encode(sig.r));
+        const ss = tlv.encode(0x02, int.encode(sig.s));
+        const seq = rs + ss;
+        return tlv.encode(0x30, seq);
+    },
+};
+Object.freeze(DER._tlv);
+Object.freeze(DER._int);
+Object.freeze(DER);
+// Be friendly to bad ECMAScript parsers by not using bigint literals
+// prettier-ignore
+const _0n = /* @__PURE__ */ BigInt(0), _1n = /* @__PURE__ */ BigInt(1), _2n$1 = /* @__PURE__ */ BigInt(2), _3n = /* @__PURE__ */ BigInt(3), _4n = /* @__PURE__ */ BigInt(4);
+/**
+ * Creates weierstrass Point constructor, based on specified curve options.
+ *
+ * See {@link WeierstrassOpts}.
+ * @param params - Curve parameters. See {@link WeierstrassOpts}.
+ * @param extraOpts - Optional helpers and overrides. See {@link WeierstrassExtraOpts}.
+ * @returns Weierstrass point constructor.
+ * @throws If the curve parameters, overrides, or point codecs are invalid. {@link Error}
+ *
+ * @example
+ * Construct a point type from explicit Weierstrass curve parameters.
+ *
+ * ```js
+ * const opts = {
+ *   p: 0xfffffffffffffffffffffffffffffffeffffac73n,
+ *   n: 0x100000000000000000001b8fa16dfab9aca16b6b3n,
+ *   h: 1n,
+ *   a: 0n,
+ *   b: 7n,
+ *   Gx: 0x3b4c382ce37aa192a4019e763036f4f5dd4d7ebbn,
+ *   Gy: 0x938cf935318fdced6bc28286531733c3f03c4feen,
+ * };
+ * const secp160k1_Point = weierstrass(opts);
+ * ```
+ */
+function weierstrass(params, extraOpts = {}) {
+    const validated = createCurveFields('weierstrass', params, extraOpts);
+    const Fp = validated.Fp;
+    const Fn = validated.Fn;
+    let CURVE = validated.CURVE;
+    const { h: cofactor, n: CURVE_ORDER } = CURVE;
+    validateObject(extraOpts, {}, {
+        allowInfinityPoint: 'boolean',
+        clearCofactor: 'function',
+        isTorsionFree: 'function',
+        fromBytes: 'function',
+        toBytes: 'function',
+        endo: 'object',
+    });
+    // Snapshot constructor-time flags whose later mutation would otherwise change
+    // validity semantics of an already-built point type.
+    const { endo, allowInfinityPoint } = extraOpts;
+    if (endo) {
+        // validateObject(endo, { beta: 'bigint', splitScalar: 'function' });
+        if (!Fp.is0(CURVE.a) || typeof endo.beta !== 'bigint' || !Array.isArray(endo.basises)) {
+            throw new Error('invalid endo: expected "beta": bigint and "basises": array');
+        }
+    }
+    const lengths = getWLengths(Fp, Fn);
+    function assertCompressionIsSupported() {
+        if (!Fp.isOdd)
+            throw new Error('compression is not supported: Field does not have .isOdd()');
+    }
+    // Implements IEEE P1363 point encoding
+    function pointToBytes(_c, point, isCompressed) {
+        // SEC 1 v2.0 §2.3.3 encodes infinity as the single octet 0x00. Only curves
+        // that opt into infinity as a public point value should expose that byte form.
+        if (allowInfinityPoint && point.is0())
+            return Uint8Array.of(0);
+        const { x, y } = point.toAffine();
+        const bx = Fp.toBytes(x);
+        abool(isCompressed, 'isCompressed');
+        if (isCompressed) {
+            assertCompressionIsSupported();
+            const hasEvenY = !Fp.isOdd(y);
+            return concatBytes(pprefix(hasEvenY), bx);
+        }
+        else {
+            return concatBytes(Uint8Array.of(0x04), bx, Fp.toBytes(y));
+        }
+    }
+    function pointFromBytes(bytes) {
+        abytes(bytes, undefined, 'Point');
+        const { publicKey: comp, publicKeyUncompressed: uncomp } = lengths; // e.g. for 32-byte: 33, 65
+        const length = bytes.length;
+        const head = bytes[0];
+        const tail = bytes.subarray(1);
+        if (allowInfinityPoint && length === 1 && head === 0x00)
+            return { x: Fp.ZERO, y: Fp.ZERO };
+        // SEC 1 v2.0 §2.3.4 decodes 0x00 as infinity, but §3.2.2 public-key validation
+        // rejects infinity. We therefore keep 0x00 rejected by default because callers
+        // reuse this parser as the strict public-key boundary, and only admit it when
+        // the curve explicitly opts into infinity as a public point value. secp256k1
+        // crosstests show OpenSSL raw point codecs accept 0x00 too.
+        // No actual validation is done here: use .assertValidity()
+        if (length === comp && (head === 0x02 || head === 0x03)) {
+            const x = Fp.fromBytes(tail);
+            if (!Fp.isValid(x))
+                throw new Error('bad point: is not on curve, wrong x');
+            const y2 = weierstrassEquation(x); // y² = x³ + ax + b
+            let y;
+            try {
+                y = Fp.sqrt(y2); // y = y² ^ (p+1)/4
+            }
+            catch (sqrtError) {
+                const err = sqrtError instanceof Error ? ': ' + sqrtError.message : '';
+                throw new Error('bad point: is not on curve, sqrt error' + err);
+            }
+            assertCompressionIsSupported();
+            const evenY = Fp.isOdd(y);
+            const evenH = (head & 1) === 1; // ECDSA-specific
+            if (evenH !== evenY)
+                y = Fp.neg(y);
+            return { x, y };
+        }
+        else if (length === uncomp && head === 0x04) {
+            // TODO: more checks
+            const L = Fp.BYTES;
+            const x = Fp.fromBytes(tail.subarray(0, L));
+            const y = Fp.fromBytes(tail.subarray(L, L * 2));
+            if (!isValidXY(x, y))
+                throw new Error('bad point: is not on curve');
+            return { x, y };
+        }
+        else {
+            throw new Error(`bad point: got length ${length}, expected compressed=${comp} or uncompressed=${uncomp}`);
+        }
+    }
+    const encodePoint = extraOpts.toBytes === undefined ? pointToBytes : extraOpts.toBytes;
+    const decodePoint = extraOpts.fromBytes === undefined ? pointFromBytes : extraOpts.fromBytes;
+    function weierstrassEquation(x) {
+        const x2 = Fp.sqr(x); // x * x
+        const x3 = Fp.mul(x2, x); // x² * x
+        return Fp.add(Fp.add(x3, Fp.mul(x, CURVE.a)), CURVE.b); // x³ + a * x + b
+    }
+    // TODO: move top-level
+    /** Checks whether equation holds for given x, y: y² == x³ + ax + b */
+    function isValidXY(x, y) {
+        const left = Fp.sqr(y); // y²
+        const right = weierstrassEquation(x); // x³ + ax + b
+        return Fp.eql(left, right);
+    }
+    // Keep constructor-time generator validation cheap: callers are responsible for supplying the
+    // correct prime-order base point, while eager subgroup checks here would slow heavy module imports.
+    // Test 1: equation y² = x³ + ax + b should work for generator point.
+    if (!isValidXY(CURVE.Gx, CURVE.Gy))
+        throw new Error('bad curve params: generator point');
+    // Test 2: discriminant Δ part should be non-zero: 4a³ + 27b² != 0.
+    // Guarantees curve is genus-1, smooth (non-singular).
+    const _4a3 = Fp.mul(Fp.pow(CURVE.a, _3n), _4n);
+    const _27b2 = Fp.mul(Fp.sqr(CURVE.b), BigInt(27));
+    if (Fp.is0(Fp.add(_4a3, _27b2)))
+        throw new Error('bad curve params: a or b');
+    /** Asserts coordinate is valid: 0 <= n < Fp.ORDER. */
+    function acoord(title, n, banZero = false) {
+        if (!Fp.isValid(n) || (banZero && Fp.is0(n)))
+            throw new Error(`bad point coordinate ${title}`);
+        return n;
+    }
+    function aprjpoint(other) {
+        if (!(other instanceof Point))
+            throw new Error('Weierstrass Point expected');
+    }
+    function splitEndoScalarN(k) {
+        if (!endo || !endo.basises)
+            throw new Error('no endo');
+        return _splitEndoScalar(k, endo.basises, Fn.ORDER);
+    }
+    function finishEndo(endoBeta, k1p, k2p, k1neg, k2neg) {
+        k2p = new Point(Fp.mul(k2p.X, endoBeta), k2p.Y, k2p.Z);
+        k1p = negateCt(k1neg, k1p);
+        k2p = negateCt(k2neg, k2p);
+        return k1p.add(k2p);
+    }
+    /**
+     * Projective Point works in 3d / projective (homogeneous) coordinates:(X, Y, Z) ∋ (x=X/Z, y=Y/Z).
+     * Default Point works in 2d / affine coordinates: (x, y).
+     * We're doing calculations in projective, because its operations don't require costly inversion.
+     */
+    class Point {
+        // base / generator point
+        static BASE = new Point(CURVE.Gx, CURVE.Gy, Fp.ONE);
+        // zero / infinity / identity point
+        static ZERO = new Point(Fp.ZERO, Fp.ONE, Fp.ZERO); // 0, 1, 0
+        // math field
+        static Fp = Fp;
+        // scalar field
+        static Fn = Fn;
+        X;
+        Y;
+        Z;
+        /** Does NOT validate if the point is valid. Use `.assertValidity()`. */
+        constructor(X, Y, Z) {
+            this.X = acoord('x', X);
+            // This is not just about ZERO / infinity: ambient curves can have real
+            // finite points with y=0. Those points are 2-torsion, so they cannot lie
+            // in the odd prime-order subgroups this point type is meant to represent.
+            this.Y = acoord('y', Y, true);
+            this.Z = acoord('z', Z);
+            Object.freeze(this);
+        }
+        static CURVE() {
+            return CURVE;
+        }
+        /** Does NOT validate if the point is valid. Use `.assertValidity()`. */
+        static fromAffine(p) {
+            const { x, y } = p || {};
+            if (!p || !Fp.isValid(x) || !Fp.isValid(y))
+                throw new Error('invalid affine point');
+            if (p instanceof Point)
+                throw new Error('projective point not allowed');
+            // (0, 0) would've produced (0, 0, 1) - instead, we need (0, 1, 0)
+            if (Fp.is0(x) && Fp.is0(y))
+                return Point.ZERO;
+            return new Point(x, y, Fp.ONE);
+        }
+        static fromBytes(bytes) {
+            const P = Point.fromAffine(decodePoint(abytes(bytes, undefined, 'point')));
+            P.assertValidity();
+            return P;
+        }
+        static fromHex(hex) {
+            return Point.fromBytes(hexToBytes(hex));
+        }
+        get x() {
+            return this.toAffine().x;
+        }
+        get y() {
+            return this.toAffine().y;
+        }
+        /**
+         *
+         * @param windowSize
+         * @param isLazy - true will defer table computation until the first multiplication
+         * @returns
+         */
+        precompute(windowSize = 8, isLazy = true) {
+            wnaf.createCache(this, windowSize);
+            if (!isLazy)
+                this.multiply(_3n); // random number
+            return this;
+        }
+        // TODO: return `this`
+        /** A point on curve is valid if it conforms to equation. */
+        assertValidity() {
+            const p = this;
+            if (p.is0()) {
+                // (0, 1, 0) aka ZERO is invalid in most contexts.
+                // In BLS, ZERO can be serialized, so we allow it.
+                // Keep the accepted infinity encoding canonical: projective-equivalent (X, Y, 0) points
+                // like (1, 1, 0) compare equal to ZERO, but only (0, 1, 0) should pass this guard.
+                if (extraOpts.allowInfinityPoint && Fp.is0(p.X) && Fp.eql(p.Y, Fp.ONE) && Fp.is0(p.Z))
+                    return;
+                throw new Error('bad point: ZERO');
+            }
+            // Some 3rd-party test vectors require different wording between here & `fromCompressedHex`
+            const { x, y } = p.toAffine();
+            if (!Fp.isValid(x) || !Fp.isValid(y))
+                throw new Error('bad point: x or y not field elements');
+            if (!isValidXY(x, y))
+                throw new Error('bad point: equation left != right');
+            if (!p.isTorsionFree())
+                throw new Error('bad point: not in prime-order subgroup');
+        }
+        hasEvenY() {
+            const { y } = this.toAffine();
+            if (!Fp.isOdd)
+                throw new Error("Field doesn't support isOdd");
+            return !Fp.isOdd(y);
+        }
+        /** Compare one point to another. */
+        equals(other) {
+            aprjpoint(other);
+            const { X: X1, Y: Y1, Z: Z1 } = this;
+            const { X: X2, Y: Y2, Z: Z2 } = other;
+            const U1 = Fp.eql(Fp.mul(X1, Z2), Fp.mul(X2, Z1));
+            const U2 = Fp.eql(Fp.mul(Y1, Z2), Fp.mul(Y2, Z1));
+            return U1 && U2;
+        }
+        /** Flips point to one corresponding to (x, -y) in Affine coordinates. */
+        negate() {
+            return new Point(this.X, Fp.neg(this.Y), this.Z);
+        }
+        // Renes-Costello-Batina exception-free doubling formula.
+        // There is 30% faster Jacobian formula, but it is not complete.
+        // https://eprint.iacr.org/2015/1060, algorithm 3
+        // Cost: 8M + 3S + 3*a + 2*b3 + 15add.
+        double() {
+            const { a, b } = CURVE;
+            const b3 = Fp.mul(b, _3n);
+            const { X: X1, Y: Y1, Z: Z1 } = this;
+            let X3 = Fp.ZERO, Y3 = Fp.ZERO, Z3 = Fp.ZERO; // prettier-ignore
+            let t0 = Fp.mul(X1, X1); // step 1
+            let t1 = Fp.mul(Y1, Y1);
+            let t2 = Fp.mul(Z1, Z1);
+            let t3 = Fp.mul(X1, Y1);
+            t3 = Fp.add(t3, t3); // step 5
+            Z3 = Fp.mul(X1, Z1);
+            Z3 = Fp.add(Z3, Z3);
+            X3 = Fp.mul(a, Z3);
+            Y3 = Fp.mul(b3, t2);
+            Y3 = Fp.add(X3, Y3); // step 10
+            X3 = Fp.sub(t1, Y3);
+            Y3 = Fp.add(t1, Y3);
+            Y3 = Fp.mul(X3, Y3);
+            X3 = Fp.mul(t3, X3);
+            Z3 = Fp.mul(b3, Z3); // step 15
+            t2 = Fp.mul(a, t2);
+            t3 = Fp.sub(t0, t2);
+            t3 = Fp.mul(a, t3);
+            t3 = Fp.add(t3, Z3);
+            Z3 = Fp.add(t0, t0); // step 20
+            t0 = Fp.add(Z3, t0);
+            t0 = Fp.add(t0, t2);
+            t0 = Fp.mul(t0, t3);
+            Y3 = Fp.add(Y3, t0);
+            t2 = Fp.mul(Y1, Z1); // step 25
+            t2 = Fp.add(t2, t2);
+            t0 = Fp.mul(t2, t3);
+            X3 = Fp.sub(X3, t0);
+            Z3 = Fp.mul(t2, t1);
+            Z3 = Fp.add(Z3, Z3); // step 30
+            Z3 = Fp.add(Z3, Z3);
+            return new Point(X3, Y3, Z3);
+        }
+        // Renes-Costello-Batina exception-free addition formula.
+        // There is 30% faster Jacobian formula, but it is not complete.
+        // https://eprint.iacr.org/2015/1060, algorithm 1
+        // Cost: 12M + 0S + 3*a + 3*b3 + 23add.
+        add(other) {
+            aprjpoint(other);
+            const { X: X1, Y: Y1, Z: Z1 } = this;
+            const { X: X2, Y: Y2, Z: Z2 } = other;
+            let X3 = Fp.ZERO, Y3 = Fp.ZERO, Z3 = Fp.ZERO; // prettier-ignore
+            const a = CURVE.a;
+            const b3 = Fp.mul(CURVE.b, _3n);
+            let t0 = Fp.mul(X1, X2); // step 1
+            let t1 = Fp.mul(Y1, Y2);
+            let t2 = Fp.mul(Z1, Z2);
+            let t3 = Fp.add(X1, Y1);
+            let t4 = Fp.add(X2, Y2); // step 5
+            t3 = Fp.mul(t3, t4);
+            t4 = Fp.add(t0, t1);
+            t3 = Fp.sub(t3, t4);
+            t4 = Fp.add(X1, Z1);
+            let t5 = Fp.add(X2, Z2); // step 10
+            t4 = Fp.mul(t4, t5);
+            t5 = Fp.add(t0, t2);
+            t4 = Fp.sub(t4, t5);
+            t5 = Fp.add(Y1, Z1);
+            X3 = Fp.add(Y2, Z2); // step 15
+            t5 = Fp.mul(t5, X3);
+            X3 = Fp.add(t1, t2);
+            t5 = Fp.sub(t5, X3);
+            Z3 = Fp.mul(a, t4);
+            X3 = Fp.mul(b3, t2); // step 20
+            Z3 = Fp.add(X3, Z3);
+            X3 = Fp.sub(t1, Z3);
+            Z3 = Fp.add(t1, Z3);
+            Y3 = Fp.mul(X3, Z3);
+            t1 = Fp.add(t0, t0); // step 25
+            t1 = Fp.add(t1, t0);
+            t2 = Fp.mul(a, t2);
+            t4 = Fp.mul(b3, t4);
+            t1 = Fp.add(t1, t2);
+            t2 = Fp.sub(t0, t2); // step 30
+            t2 = Fp.mul(a, t2);
+            t4 = Fp.add(t4, t2);
+            t0 = Fp.mul(t1, t4);
+            Y3 = Fp.add(Y3, t0);
+            t0 = Fp.mul(t5, t4); // step 35
+            X3 = Fp.mul(t3, X3);
+            X3 = Fp.sub(X3, t0);
+            t0 = Fp.mul(t3, t1);
+            Z3 = Fp.mul(t5, Z3);
+            Z3 = Fp.add(Z3, t0); // step 40
+            return new Point(X3, Y3, Z3);
+        }
+        subtract(other) {
+            // Validate before calling `negate()` so wrong inputs fail with the point guard
+            // instead of leaking a foreign `negate()` error.
+            aprjpoint(other);
+            return this.add(other.negate());
+        }
+        is0() {
+            return this.equals(Point.ZERO);
+        }
+        /**
+         * Constant time multiplication.
+         * Uses wNAF method. Windowed method may be 10% faster,
+         * but takes 2x longer to generate and consumes 2x memory.
+         * Uses precomputes when available.
+         * Uses endomorphism for Koblitz curves.
+         * @param scalar - by which the point would be multiplied
+         * @returns New point
+         */
+        multiply(scalar) {
+            const { endo } = extraOpts;
+            // Keep the subgroup-scalar contract strict instead of reducing 0 / n to ZERO.
+            // In key/signature-style callers, those values usually mean broken hash/scalar plumbing,
+            // and failing closed is safer than silently producing the identity point.
+            if (!Fn.isValidNot0(scalar))
+                throw new RangeError('invalid scalar: out of range'); // 0 is invalid
+            let point, fake; // Fake point is used to const-time mult
+            const mul = (n) => wnaf.cached(this, n, (p) => normalizeZ(Point, p));
+            /** See docs for {@link EndomorphismOpts} */
+            if (endo) {
+                const { k1neg, k1, k2neg, k2 } = splitEndoScalarN(scalar);
+                const { p: k1p, f: k1f } = mul(k1);
+                const { p: k2p, f: k2f } = mul(k2);
+                fake = k1f.add(k2f);
+                point = finishEndo(endo.beta, k1p, k2p, k1neg, k2neg);
+            }
+            else {
+                const { p, f } = mul(scalar);
+                point = p;
+                fake = f;
+            }
+            // Normalize `z` for both points, but return only real one
+            return normalizeZ(Point, [point, fake])[0];
+        }
+        /**
+         * Non-constant-time multiplication. Uses double-and-add algorithm.
+         * It's faster, but should only be used when you don't care about
+         * an exposed secret key e.g. sig verification, which works over *public* keys.
+         */
+        multiplyUnsafe(scalar) {
+            const { endo } = extraOpts;
+            const p = this;
+            const sc = scalar;
+            // Public-scalar callers may need 0, but n and larger values stay rejected here too.
+            // Reducing them mod n would turn bad caller input into an accidental identity point.
+            if (!Fn.isValid(sc))
+                throw new RangeError('invalid scalar: out of range'); // 0 is valid
+            if (sc === _0n || p.is0())
+                return Point.ZERO; // 0
+            if (sc === _1n)
+                return p; // 1
+            if (wnaf.hasCache(this))
+                return this.multiply(sc); // precomputes
+            // We don't have method for double scalar multiplication (aP + bQ):
+            // Even with using Strauss-Shamir trick, it's 35% slower than naïve mul+add.
+            if (endo) {
+                const { k1neg, k1, k2neg, k2 } = splitEndoScalarN(sc);
+                const { p1, p2 } = mulEndoUnsafe(Point, p, k1, k2); // 30% faster vs wnaf.unsafe
+                return finishEndo(endo.beta, p1, p2, k1neg, k2neg);
+            }
+            else {
+                return wnaf.unsafe(p, sc);
+            }
+        }
+        /**
+         * Converts Projective point to affine (x, y) coordinates.
+         * (X, Y, Z) ∋ (x=X/Z, y=Y/Z).
+         * @param invertedZ - Z^-1 (inverted zero) - optional, precomputation is useful for invertBatch
+         */
+        toAffine(invertedZ) {
+            const p = this;
+            let iz = invertedZ;
+            const { X, Y, Z } = p;
+            // Fast-path for normalized points
+            if (Fp.eql(Z, Fp.ONE))
+                return { x: X, y: Y };
+            const is0 = p.is0();
+            // If invZ was 0, we return zero point. However we still want to execute
+            // all operations, so we replace invZ with a random number, 1.
+            if (iz == null)
+                iz = is0 ? Fp.ONE : Fp.inv(Z);
+            const x = Fp.mul(X, iz);
+            const y = Fp.mul(Y, iz);
+            const zz = Fp.mul(Z, iz);
+            if (is0)
+                return { x: Fp.ZERO, y: Fp.ZERO };
+            if (!Fp.eql(zz, Fp.ONE))
+                throw new Error('invZ was invalid');
+            return { x, y };
+        }
+        /**
+         * Checks whether Point is free of torsion elements (is in prime subgroup).
+         * Always torsion-free for cofactor=1 curves.
+         */
+        isTorsionFree() {
+            const { isTorsionFree } = extraOpts;
+            if (cofactor === _1n)
+                return true;
+            if (isTorsionFree)
+                return isTorsionFree(Point, this);
+            return wnaf.unsafe(this, CURVE_ORDER).is0();
+        }
+        clearCofactor() {
+            const { clearCofactor } = extraOpts;
+            if (cofactor === _1n)
+                return this; // Fast-path
+            if (clearCofactor)
+                return clearCofactor(Point, this);
+            // Default fallback assumes the cofactor fits the usual subgroup-scalar
+            // multiplyUnsafe() contract. Curves with larger / structured cofactors
+            // should define a clearCofactor override anyway (e.g. psi/Frobenius maps).
+            return this.multiplyUnsafe(cofactor);
+        }
+        isSmallOrder() {
+            if (cofactor === _1n)
+                return this.is0(); // Fast-path
+            return this.clearCofactor().is0();
+        }
+        toBytes(isCompressed = true) {
+            abool(isCompressed, 'isCompressed');
+            // Same policy as pointFromBytes(): keep ZERO out of the default byte surface because
+            // callers use these encodings as public keys, where SEC 1 validation rejects infinity.
+            this.assertValidity();
+            return encodePoint(Point, this, isCompressed);
+        }
+        toHex(isCompressed = true) {
+            return bytesToHex(this.toBytes(isCompressed));
+        }
+        toString() {
+            return `<Point ${this.is0() ? 'ZERO' : this.toHex()}>`;
+        }
+    }
+    const bits = Fn.BITS;
+    const wnaf = new wNAF(Point, extraOpts.endo ? Math.ceil(bits / 2) : bits);
+    // Tiny toy curves can have scalar fields narrower than 8 bits. Skip the
+    // eager W=8 cache there instead of rejecting an otherwise valid constructor.
+    if (bits >= 8)
+        Point.BASE.precompute(8); // Enable precomputes. Slows down first publicKey computation by 20ms.
+    Object.freeze(Point.prototype);
+    Object.freeze(Point);
+    return Point;
+}
+// Points start with byte 0x02 when y is even; otherwise 0x03
+function pprefix(hasEvenY) {
+    return Uint8Array.of(hasEvenY ? 0x02 : 0x03);
+}
+function getWLengths(Fp, Fn) {
+    return {
+        secretKey: Fn.BYTES,
+        publicKey: 1 + Fp.BYTES,
+        publicKeyUncompressed: 1 + 2 * Fp.BYTES,
+        publicKeyHasPrefix: true,
+        // Raw compact `(r || s)` signature width; DER and recovered signatures use
+        // different lengths outside this helper.
+        signature: 2 * Fn.BYTES,
+    };
+}
+/**
+ * Sometimes users only need getPublicKey, getSharedSecret, and secret key handling.
+ * This helper ensures no signature functionality is present. Less code, smaller bundle size.
+ * @param Point - Weierstrass point constructor.
+ * @param ecdhOpts - Optional randomness helpers:
+ *   - `randomBytes` (optional): Optional RNG override.
+ * @returns ECDH helper namespace.
+ * @example
+ * Sometimes users only need getPublicKey, getSharedSecret, and secret key handling.
+ *
+ * ```ts
+ * import { ecdh } from '@noble/curves/abstract/weierstrass.js';
+ * import { p256 } from '@noble/curves/nist.js';
+ * const dh = ecdh(p256.Point);
+ * const alice = dh.keygen();
+ * const shared = dh.getSharedSecret(alice.secretKey, alice.publicKey);
+ * ```
+ */
+function ecdh(Point, ecdhOpts = {}) {
+    const { Fn } = Point;
+    const randomBytes_ = ecdhOpts.randomBytes === undefined ? randomBytes : ecdhOpts.randomBytes;
+    // Keep the advertised seed length aligned with mapHashToField(), which keeps a hard 16-byte
+    // minimum even on toy curves.
+    const lengths = Object.assign(getWLengths(Point.Fp, Fn), {
+        seed: Math.max(getMinHashLength(Fn.ORDER), 16),
+    });
+    function isValidSecretKey(secretKey) {
+        try {
+            const num = Fn.fromBytes(secretKey);
+            return Fn.isValidNot0(num);
+        }
+        catch (error) {
+            return false;
+        }
+    }
+    function isValidPublicKey(publicKey, isCompressed) {
+        const { publicKey: comp, publicKeyUncompressed } = lengths;
+        try {
+            const l = publicKey.length;
+            if (isCompressed === true && l !== comp)
+                return false;
+            if (isCompressed === false && l !== publicKeyUncompressed)
+                return false;
+            return !!Point.fromBytes(publicKey);
+        }
+        catch (error) {
+            return false;
+        }
+    }
+    /**
+     * Produces cryptographically secure secret key from random of size
+     * (groupLen + ceil(groupLen / 2)) with modulo bias being negligible.
+     */
+    function randomSecretKey(seed) {
+        seed = seed === undefined ? randomBytes_(lengths.seed) : seed;
+        return mapHashToField(abytes(seed, lengths.seed, 'seed'), Fn.ORDER);
+    }
+    /**
+     * Computes public key for a secret key. Checks for validity of the secret key.
+     * @param isCompressed - whether to return compact (default), or full key
+     * @returns Public key, full when isCompressed=false; short when isCompressed=true
+     */
+    function getPublicKey(secretKey, isCompressed = true) {
+        return Point.BASE.multiply(Fn.fromBytes(secretKey)).toBytes(isCompressed);
+    }
+    /**
+     * Quick and dirty check for item being public key. Does not validate hex, or being on-curve.
+     */
+    function isProbPub(item) {
+        const { secretKey, publicKey, publicKeyUncompressed } = lengths;
+        const allowedLengths = Fn._lengths;
+        if (!isBytes(item))
+            return undefined;
+        const l = abytes(item, undefined, 'key').length;
+        const isPub = l === publicKey || l === publicKeyUncompressed;
+        const isSec = l === secretKey || !!allowedLengths?.includes(l);
+        // P-521 accepts both 65- and 66-byte secret keys, so overlapping lengths stay ambiguous.
+        if (isPub && isSec)
+            return undefined;
+        return isPub;
+    }
+    /**
+     * ECDH (Elliptic Curve Diffie Hellman).
+     * Computes encoded shared point from secret key A and public key B.
+     * Checks: 1) secret key validity 2) shared key is on-curve.
+     * Does NOT hash the result or expose the SEC 1 x-coordinate-only `z`.
+     * Returns the encoded shared point on purpose: callers that need `x_P`
+     * can derive it from the encoded point, but `x_P` alone cannot recover the
+     * point/parity back.
+     * This helper only exposes the fully validated public-key path, not cofactor DH.
+     * @param isCompressed - whether to return compact (default), or full key
+     * @returns shared point encoding
+     */
+    function getSharedSecret(secretKeyA, publicKeyB, isCompressed = true) {
+        if (isProbPub(secretKeyA) === true)
+            throw new Error('first arg must be private key');
+        if (isProbPub(publicKeyB) === false)
+            throw new Error('second arg must be public key');
+        const s = Fn.fromBytes(secretKeyA);
+        const b = Point.fromBytes(publicKeyB); // checks for being on-curve
+        return b.multiply(s).toBytes(isCompressed);
+    }
+    const utils = {
+        isValidSecretKey,
+        isValidPublicKey,
+        randomSecretKey,
+    };
+    const keygen = createKeygen(randomSecretKey, getPublicKey);
+    Object.freeze(utils);
+    Object.freeze(lengths);
+    return Object.freeze({ getPublicKey, getSharedSecret, keygen, Point, utils, lengths });
+}
+/**
+ * Creates ECDSA signing interface for given elliptic curve `Point` and `hash` function.
+ *
+ * @param Point - created using {@link weierstrass} function
+ * @param hash - used for 1) message prehash-ing 2) k generation in `sign`, using hmac_drbg(hash)
+ * @param ecdsaOpts - rarely needed, see {@link ECDSAOpts}:
+ *   - `lowS`: Default low-S policy.
+ *   - `hmac`: HMAC implementation used by RFC6979 DRBG.
+ *   - `randomBytes`: Optional RNG override.
+ *   - `bits2int`: Optional hash-to-int conversion override.
+ *   - `bits2int_modN`: Optional hash-to-int-mod-n conversion override.
+ *
+ * @returns ECDSA helper namespace.
+ * @example
+ * Create an ECDSA signer/verifier bundle for one curve implementation.
+ *
+ * ```ts
+ * import { ecdsa } from '@noble/curves/abstract/weierstrass.js';
+ * import { p256 } from '@noble/curves/nist.js';
+ * import { sha256 } from '@noble/hashes/sha2.js';
+ * const p256ecdsa = ecdsa(p256.Point, sha256);
+ * const { secretKey, publicKey } = p256ecdsa.keygen();
+ * const msg = new TextEncoder().encode('hello noble');
+ * const sig = p256ecdsa.sign(msg, secretKey);
+ * const isValid = p256ecdsa.verify(sig, msg, publicKey);
+ * ```
+ */
+function ecdsa(Point, hash, ecdsaOpts = {}) {
+    // Custom hash / bits2int hooks are treated as pure functions over validated caller-owned bytes.
+    const hash_ = hash;
+    ahash(hash_);
+    validateObject(ecdsaOpts, {}, {
+        hmac: 'function',
+        lowS: 'boolean',
+        randomBytes: 'function',
+        bits2int: 'function',
+        bits2int_modN: 'function',
+    });
+    ecdsaOpts = Object.assign({}, ecdsaOpts);
+    const randomBytes$1 = ecdsaOpts.randomBytes === undefined ? randomBytes : ecdsaOpts.randomBytes;
+    const hmac$1 = ecdsaOpts.hmac === undefined
+        ? (key, msg) => hmac(hash_, key, msg)
+        : ecdsaOpts.hmac;
+    const { Fp, Fn } = Point;
+    const { ORDER: CURVE_ORDER, BITS: fnBits } = Fn;
+    const { keygen, getPublicKey, getSharedSecret, utils, lengths } = ecdh(Point, ecdsaOpts);
+    const defaultSigOpts = {
+        prehash: true,
+        lowS: typeof ecdsaOpts.lowS === 'boolean' ? ecdsaOpts.lowS : true,
+        format: 'compact',
+        extraEntropy: false,
+    };
+    // SEC 1 4.1.6 public-key recovery tries x = r + jn for j = 0..h. Our recovered-signature
+    // format only stores one overflow bit, so it can only distinguish q.x = r from q.x = r + n.
+    // A third lift would have the form q.x = r + 2n. Since valid ECDSA r is in 1..n-1, the
+    // smallest such lift is 1 + 2n, not 2n.
+    const hasLargeRecoveryLifts = CURVE_ORDER * _2n$1 + _1n < Fp.ORDER;
+    function isBiggerThanHalfOrder(number) {
+        const HALF = CURVE_ORDER >> _1n;
+        return number > HALF;
+    }
+    function validateRS(title, num) {
+        if (!Fn.isValidNot0(num))
+            throw new Error(`invalid signature ${title}: out of range 1..Point.Fn.ORDER`);
+        return num;
+    }
+    function assertRecoverableCurve() {
+        // ECDSA recovery only supports curves where the current recovery id can distinguish
+        // q.x = r and q.x = r + n; larger lifts may need additional `r + n*i` branches.
+        // SEC 1 4.1.6 recovers candidates via x = r + jn, but this format only encodes j = 0 or 1.
+        // The next possible candidate is q.x = r + 2n, and its smallest valid value is 1 + 2n.
+        // To easily get i, we either need to:
+        // a. increase amount of valid recid values (4, 5...); OR
+        // b. prohibit recovered signatures for those curves.
+        if (hasLargeRecoveryLifts)
+            throw new Error('"recovered" sig type is not supported for cofactor >2 curves');
+    }
+    function validateSigLength(bytes, format) {
+        validateSigFormat(format);
+        const size = lengths.signature;
+        const sizer = format === 'compact' ? size : format === 'recovered' ? size + 1 : undefined;
+        return abytes(bytes, sizer);
+    }
+    /**
+     * ECDSA signature with its (r, s) properties. Supports compact, recovered & DER representations.
+     */
+    class Signature {
+        r;
+        s;
+        recovery;
+        constructor(r, s, recovery) {
+            this.r = validateRS('r', r); // r in [1..N-1];
+            this.s = validateRS('s', s); // s in [1..N-1];
+            if (recovery != null) {
+                assertRecoverableCurve();
+                if (![0, 1, 2, 3].includes(recovery))
+                    throw new Error('invalid recovery id');
+                this.recovery = recovery;
+            }
+            Object.freeze(this);
+        }
+        static fromBytes(bytes, format = defaultSigOpts.format) {
+            validateSigLength(bytes, format);
+            let recid;
+            if (format === 'der') {
+                const { r, s } = DER.toSig(abytes(bytes));
+                return new Signature(r, s);
+            }
+            if (format === 'recovered') {
+                recid = bytes[0];
+                format = 'compact';
+                bytes = bytes.subarray(1);
+            }
+            const L = lengths.signature / 2;
+            const r = bytes.subarray(0, L);
+            const s = bytes.subarray(L, L * 2);
+            return new Signature(Fn.fromBytes(r), Fn.fromBytes(s), recid);
+        }
+        static fromHex(hex, format) {
+            return this.fromBytes(hexToBytes(hex), format);
+        }
+        assertRecovery() {
+            const { recovery } = this;
+            if (recovery == null)
+                throw new Error('invalid recovery id: must be present');
+            return recovery;
+        }
+        addRecoveryBit(recovery) {
+            return new Signature(this.r, this.s, recovery);
+        }
+        // Unlike the top-level helper below, this method expects a digest that has
+        // already been hashed to the curve's message representative.
+        recoverPublicKey(messageHash) {
+            const { r, s } = this;
+            const recovery = this.assertRecovery();
+            const radj = recovery === 2 || recovery === 3 ? r + CURVE_ORDER : r;
+            if (!Fp.isValid(radj))
+                throw new Error('invalid recovery id: sig.r+curve.n != R.x');
+            const x = Fp.toBytes(radj);
+            const R = Point.fromBytes(concatBytes(pprefix((recovery & 1) === 0), x));
+            const ir = Fn.inv(radj); // r^-1
+            const h = bits2int_modN(abytes(messageHash, undefined, 'msgHash')); // Truncate hash
+            const u1 = Fn.create(-h * ir); // -hr^-1
+            const u2 = Fn.create(s * ir); // sr^-1
+            // (sr^-1)R-(hr^-1)G = -(hr^-1)G + (sr^-1). unsafe is fine: there is no private data.
+            const Q = Point.BASE.multiplyUnsafe(u1).add(R.multiplyUnsafe(u2));
+            if (Q.is0())
+                throw new Error('invalid recovery: point at infinify');
+            Q.assertValidity();
+            return Q;
+        }
+        // Signatures should be low-s, to prevent malleability.
+        hasHighS() {
+            return isBiggerThanHalfOrder(this.s);
+        }
+        toBytes(format = defaultSigOpts.format) {
+            validateSigFormat(format);
+            if (format === 'der')
+                return hexToBytes(DER.hexFromSig(this));
+            const { r, s } = this;
+            const rb = Fn.toBytes(r);
+            const sb = Fn.toBytes(s);
+            if (format === 'recovered') {
+                assertRecoverableCurve();
+                return concatBytes(Uint8Array.of(this.assertRecovery()), rb, sb);
+            }
+            return concatBytes(rb, sb);
+        }
+        toHex(format) {
+            return bytesToHex(this.toBytes(format));
+        }
+    }
+    Object.freeze(Signature.prototype);
+    Object.freeze(Signature);
+    // RFC6979: ensure ECDSA msg is X bytes and < N. RFC suggests optional truncating via bits2octets.
+    // FIPS 186-4 4.6 suggests the leftmost min(nBitLen, outLen) bits, which matches bits2int.
+    // bits2int can produce res>N, we can do mod(res, N) since the bitLen is the same.
+    // int2octets can't be used; pads small msgs with 0: unacceptatble for trunc as per RFC vectors
+    const bits2int = ecdsaOpts.bits2int === undefined
+        ? function bits2int_def(bytes) {
+            // Our custom check "just in case", for protection against DoS
+            if (bytes.length > 8192)
+                throw new Error('input is too large');
+            // For curves with nBitLength % 8 !== 0: bits2octets(bits2octets(m)) !== bits2octets(m)
+            // for some cases, since bytes.length * 8 is not actual bitLength.
+            const num = bytesToNumberBE(bytes); // check for == u8 done here
+            const delta = bytes.length * 8 - fnBits; // truncate to nBitLength leftmost bits
+            return delta > 0 ? num >> BigInt(delta) : num;
+        }
+        : ecdsaOpts.bits2int;
+    const bits2int_modN = ecdsaOpts.bits2int_modN === undefined
+        ? function bits2int_modN_def(bytes) {
+            return Fn.create(bits2int(bytes)); // can't use bytesToNumberBE here
+        }
+        : ecdsaOpts.bits2int_modN;
+    const ORDER_MASK = bitMask(fnBits);
+    // Pads output with zero as per spec.
+    /** Converts to bytes. Checks if num in `[0..ORDER_MASK-1]` e.g.: `[0..2^256-1]`. */
+    function int2octets(num) {
+        aInRange('num < 2^' + fnBits, num, _0n, ORDER_MASK);
+        return Fn.toBytes(num);
+    }
+    function validateMsgAndHash(message, prehash) {
+        abytes(message, undefined, 'message');
+        return (prehash ? abytes(hash_(message), undefined, 'prehashed message') : message);
+    }
+    /**
+     * Steps A, D of RFC6979 3.2.
+     * Creates RFC6979 seed; converts msg/privKey to numbers.
+     * Used only in sign, not in verify.
+     *
+     * Warning: we cannot assume here that message has same amount of bytes as curve order,
+     * this will be invalid at least for P521. Also it can be bigger for P224 + SHA256.
+     */
+    function prepSig(message, secretKey, opts) {
+        const { lowS, prehash, extraEntropy } = validateSigOpts(opts, defaultSigOpts);
+        message = validateMsgAndHash(message, prehash); // RFC6979 3.2 A: h1 = H(m)
+        // We can't later call bits2octets, since nested bits2int is broken for curves
+        // with fnBits % 8 !== 0. Because of that, we unwrap it here as int2octets call.
+        // const bits2octets = (bits) => int2octets(bits2int_modN(bits))
+        const h1int = bits2int_modN(message);
+        const d = Fn.fromBytes(secretKey); // validate secret key, convert to bigint
+        if (!Fn.isValidNot0(d))
+            throw new Error('invalid private key');
+        const seedArgs = [int2octets(d), int2octets(h1int)];
+        // extraEntropy. RFC6979 3.6: additional k' (optional).
+        if (extraEntropy != null && extraEntropy !== false) {
+            // K = HMAC_K(V || 0x00 || int2octets(x) || bits2octets(h1) || k')
+            // gen random bytes OR pass as-is
+            const e = extraEntropy === true ? randomBytes$1(lengths.secretKey) : extraEntropy;
+            seedArgs.push(abytes(e, undefined, 'extraEntropy')); // check for being bytes
+        }
+        const seed = concatBytes(...seedArgs); // Step D of RFC6979 3.2
+        const m = h1int; // no need to call bits2int second time here, it is inside truncateHash!
+        // Converts signature params into point w r/s, checks result for validity.
+        // To transform k => Signature:
+        // q = k⋅G
+        // r = q.x mod n
+        // s = k^-1(m + rd) mod n
+        // Can use scalar blinding b^-1(bm + bdr) where b ∈ [1,q−1] according to
+        // https://tches.iacr.org/index.php/TCHES/article/view/7337/6509. We've decided against it:
+        // a) dependency on CSPRNG b) 15% slowdown c) doesn't really help since bigints are not CT
+        function k2sig(kBytes) {
+            // RFC 6979 Section 3.2, step 3: k = bits2int(T)
+            // Important: all mod() calls here must be done over N
+            const k = bits2int(kBytes); // Cannot use fields methods, since it is group element
+            if (!Fn.isValidNot0(k))
+                return; // Valid scalars (including k) must be in 1..N-1
+            const ik = Fn.inv(k); // k^-1 mod n
+            const q = Point.BASE.multiply(k).toAffine(); // q = k⋅G
+            const r = Fn.create(q.x); // r = q.x mod n
+            if (r === _0n)
+                return;
+            const s = Fn.create(ik * Fn.create(m + r * d)); // s = k^-1(m + rd) mod n
+            if (s === _0n)
+                return;
+            let recovery = (q.x === r ? 0 : 2) | Number(q.y & _1n); // recovery bit (2 or 3 when q.x>n)
+            let normS = s;
+            if (lowS && isBiggerThanHalfOrder(s)) {
+                normS = Fn.neg(s); // if lowS was passed, ensure s is always in the bottom half of N
+                recovery ^= 1;
+            }
+            return new Signature(r, normS, hasLargeRecoveryLifts ? undefined : recovery);
+        }
+        return { seed, k2sig };
+    }
+    /**
+     * Signs a message or message hash with a secret key.
+     * With the default `prehash: true`, raw message bytes are hashed internally;
+     * only `{ prehash: false }` expects a caller-supplied digest.
+     *
+     * ```
+     * sign(m, d) where
+     *   k = rfc6979_hmac_drbg(m, d)
+     *   (x, y) = G × k
+     *   r = x mod n
+     *   s = (m + dr) / k mod n
+     * ```
+     */
+    function sign(message, secretKey, opts = {}) {
+        const { seed, k2sig } = prepSig(message, secretKey, opts); // Steps A, D of RFC6979 3.2.
+        const drbg = createHmacDrbg(hash_.outputLen, Fn.BYTES, hmac$1);
+        const sig = drbg(seed, k2sig); // Steps B, C, D, E, F, G
+        return sig.toBytes(opts.format);
+    }
+    /**
+     * Verifies a signature against message and public key.
+     * Rejects lowS signatures by default: see {@link ECDSAVerifyOpts}.
+     * Implements section 4.1.4 from https://www.secg.org/sec1-v2.pdf:
+     *
+     * ```
+     * verify(r, s, h, P) where
+     *   u1 = hs^-1 mod n
+     *   u2 = rs^-1 mod n
+     *   R = u1⋅G + u2⋅P
+     *   mod(R.x, n) == r
+     * ```
+     */
+    function verify(signature, message, publicKey, opts = {}) {
+        const { lowS, prehash, format } = validateSigOpts(opts, defaultSigOpts);
+        publicKey = abytes(publicKey, undefined, 'publicKey');
+        message = validateMsgAndHash(message, prehash);
+        if (!isBytes(signature)) {
+            const end = signature instanceof Signature ? ', use sig.toBytes()' : '';
+            throw new Error('verify expects Uint8Array signature' + end);
+        }
+        validateSigLength(signature, format); // execute this twice because we want loud error
+        try {
+            const sig = Signature.fromBytes(signature, format);
+            const P = Point.fromBytes(publicKey);
+            if (lowS && sig.hasHighS())
+                return false;
+            const { r, s } = sig;
+            const h = bits2int_modN(message); // mod n, not mod p
+            const is = Fn.inv(s); // s^-1 mod n
+            const u1 = Fn.create(h * is); // u1 = hs^-1 mod n
+            const u2 = Fn.create(r * is); // u2 = rs^-1 mod n
+            const R = Point.BASE.multiplyUnsafe(u1).add(P.multiplyUnsafe(u2)); // u1⋅G + u2⋅P
+            if (R.is0())
+                return false;
+            const v = Fn.create(R.x); // v = r.x mod n
+            return v === r;
+        }
+        catch (e) {
+            return false;
+        }
+    }
+    function recoverPublicKey(signature, message, opts = {}) {
+        // Top-level recovery mirrors `sign()` / `verify()`: it hashes raw message
+        // bytes first unless the caller passes `{ prehash: false }`.
+        const { prehash } = validateSigOpts(opts, defaultSigOpts);
+        message = validateMsgAndHash(message, prehash);
+        return Signature.fromBytes(signature, 'recovered').recoverPublicKey(message).toBytes();
+    }
+    return Object.freeze({
+        keygen,
+        getPublicKey,
+        getSharedSecret,
+        utils,
+        lengths,
+        Point,
+        sign,
+        verify,
+        recoverPublicKey,
+        Signature,
+        hash: hash_,
+    });
+}
+
+/**
+ * SECG secp256k1. See [pdf](https://www.secg.org/sec2-v2.pdf).
+ *
+ * Belongs to Koblitz curves: it has efficiently-computable GLV endomorphism ψ,
+ * check out {@link EndomorphismOpts}. Seems to be rigid (not backdoored).
+ * @module
+ */
+/*! noble-curves - MIT License (c) 2022 Paul Miller (paulmillr.com) */
+// Seems like generator was produced from some seed:
+// `Pointk1.BASE.multiply(Pointk1.Fn.inv(2n, N)).toAffine().x`
+// // gives short x 0x3b78ce563f89a0ed9414f5aa28ad0d96d6795f9c63n
+const secp256k1_CURVE = {
+    p: BigInt('0xfffffffffffffffffffffffffffffffffffffffffffffffffffffffefffffc2f'),
+    n: BigInt('0xfffffffffffffffffffffffffffffffebaaedce6af48a03bbfd25e8cd0364141'),
+    h: BigInt(1),
+    a: BigInt(0),
+    b: BigInt(7),
+    Gx: BigInt('0x79be667ef9dcbbac55a06295ce870b07029bfcdb2dce28d959f2815b16f81798'),
+    Gy: BigInt('0x483ada7726a3c4655da4fbfc0e1108a8fd17b448a68554199c47d08ffb10d4b8'),
+};
+const secp256k1_ENDO = {
+    beta: BigInt('0x7ae96a2b657c07106e64479eac3434e99cf0497512f58995c1396c28719501ee'),
+    basises: [
+        [BigInt('0x3086d221a7d46bcde86c90e49284eb15'), -BigInt('0xe4437ed6010e88286f547fa90abfe4c3')],
+        [BigInt('0x114ca50f7a8e2f3f657c1108d9d44cfd8'), BigInt('0x3086d221a7d46bcde86c90e49284eb15')],
+    ],
+};
+const _2n = /* @__PURE__ */ BigInt(2);
+/**
+ * √n = n^((p+1)/4) for fields p = 3 mod 4. We unwrap the loop and multiply bit-by-bit.
+ * (P+1n/4n).toString(2) would produce bits [223x 1, 0, 22x 1, 4x 0, 11, 00]
+ */
+function sqrtMod(y) {
+    const P = secp256k1_CURVE.p;
+    // prettier-ignore
+    const _3n = BigInt(3), _6n = BigInt(6), _11n = BigInt(11), _22n = BigInt(22);
+    // prettier-ignore
+    const _23n = BigInt(23), _44n = BigInt(44), _88n = BigInt(88);
+    const b2 = (y * y * y) % P; // x^3, 11
+    const b3 = (b2 * b2 * y) % P; // x^7
+    const b6 = (pow2(b3, _3n, P) * b3) % P;
+    const b9 = (pow2(b6, _3n, P) * b3) % P;
+    const b11 = (pow2(b9, _2n, P) * b2) % P;
+    const b22 = (pow2(b11, _11n, P) * b11) % P;
+    const b44 = (pow2(b22, _22n, P) * b22) % P;
+    const b88 = (pow2(b44, _44n, P) * b44) % P;
+    const b176 = (pow2(b88, _88n, P) * b88) % P;
+    const b220 = (pow2(b176, _44n, P) * b44) % P;
+    const b223 = (pow2(b220, _3n, P) * b3) % P;
+    const t1 = (pow2(b223, _23n, P) * b22) % P;
+    const t2 = (pow2(t1, _6n, P) * b2) % P;
+    const root = pow2(t2, _2n, P);
+    if (!Fpk1.eql(Fpk1.sqr(root), y))
+        throw new Error('Cannot find square root');
+    return root;
+}
+const Fpk1 = Field(secp256k1_CURVE.p, { sqrt: sqrtMod });
+const Pointk1 = /* @__PURE__ */ weierstrass(secp256k1_CURVE, {
+    Fp: Fpk1,
+    endo: secp256k1_ENDO,
+});
+/**
+ * secp256k1 curve: ECDSA and ECDH methods.
+ *
+ * Uses sha256 to hash messages. To use a different hash,
+ * pass `{ prehash: false }` to sign / verify.
+ *
+ * @example
+ * Generate one secp256k1 keypair, sign a message, and verify it.
+ *
+ * ```js
+ * import { secp256k1 } from '@noble/curves/secp256k1.js';
+ * const { secretKey, publicKey } = secp256k1.keygen();
+ * // const publicKey = secp256k1.getPublicKey(secretKey);
+ * const msg = new TextEncoder().encode('hello noble');
+ * const sig = secp256k1.sign(msg, secretKey);
+ * const isValid = secp256k1.verify(sig, msg, publicKey);
+ * // const sigKeccak = secp256k1.sign(keccak256(msg), secretKey, { prehash: false });
+ * ```
+ */
+const secp256k1 = /* @__PURE__ */ ecdsa(Pointk1, sha256);
+
+function canonicalize (object, seen = new Set()) {
+  if (typeof object === 'number' && isNaN(object)) {
+    throw new Error('NaN is not allowed');
+  }
+
+  if (typeof object === 'number' && !isFinite(object)) {
+    throw new Error('Infinity is not allowed');
+  }
+
+  if (object === null || typeof object !== 'object') {
+    return JSON.stringify(object);
+  }
+
+  if (typeof object.toJSON === 'function') {
+    if (seen.has(object)) {
+      throw new Error('Circular reference detected');
+    }
+    seen.add(object);
+    const result = canonicalize(object.toJSON(), seen);
+    seen.delete(object);
+    return result;
+  }
+
+  if (seen.has(object)) {
+    throw new Error('Circular reference detected');
+  }
+  seen.add(object);
+
+  let result;
+  if (Array.isArray(object)) {
+    const values = object.map((cv) => {
+      const value = cv === undefined || typeof cv === 'symbol' ? null : cv;
+      return canonicalize(value, seen);
+    });
+    result = `[${values.join(',')}]`;
+  } else {
+    const parts = [];
+    for (const key of Object.keys(object).sort()) {
+      if (object[key] === undefined || typeof object[key] === 'symbol') {
+        continue;
+      }
+      parts.push(`${canonicalize(key)}:${canonicalize(object[key], seen)}`);
+    }
+    result = `{${parts.join(',')}}`;
+  }
+
+  seen.delete(object);
+  return result;
+}
+
+/**
+ * Operator-signed trust-manifest distribution.
+ *
+ * The SDK's attestation check (see `TrustAnchor`) needs a trusted source
+ * for the expected cluster identity — the set of node peer IDs and the
+ * allow-listed RTMR1/RTMR3 image measurements (RTMR1 is the real rootfs
+ * signal, RTMR3 is kept for backward compatibility — see `TrustAnchor`'s
+ * own doc comment). Without a trusted source, every
+ * caller has to hand-pin those values or fall back to
+ * `{ unsafe_trust_server: true }`, which disables verification entirely.
+ *
+ * A trust manifest is a small JSON document published by the cluster
+ * operator to a public URL (a GCS bucket) and signed with the operator's
+ * secp256k1 key. Its public half is baked into the SDK per environment
+ * (`OPERATOR_PUBLIC_KEYS`). `fetchTrustedManifest(env)` fetches the
+ * manifest, verifies the signature against the pinned operator key, and
+ * returns a {@link TrustAnchor} that plugs straight into
+ * `T3nClientConfig.trustAnchor`.
+ *
+ * Trust model: the operator key is the root of trust. A network attacker
+ * can serve any bytes at the manifest URL, but cannot forge a signature
+ * for a manifest naming their own node identities/measurements without
+ * the operator's private key. The manifest only ever *narrows* trust to
+ * the real cluster — it never widens what the on-quote RTMR1/RTMR3
+ * pinning and peer-id set-equality already enforce inside the WASM
+ * verifier.
+ */
+/**
+ * Deterministic bytes the operator signs and the client verifies: the
+ * SHA-256 of the RFC 8785 (JCS) canonical JSON of the payload. Sharing
+ * this one helper between the signer and the verifier is what guarantees
+ * both sides hash identical bytes regardless of key ordering or
+ * whitespace.
+ */
+function canonicalManifestBytes(payload) {
+    const canonical = canonicalize(payload);
+    if (typeof canonical !== "string") {
+        throw new Error("trust manifest payload is not canonicalizable JSON");
+    }
+    return sha256(new TextEncoder().encode(canonical));
+}
+
+/**
+ * Sign a trust manifest with the operator's secp256k1 private key.
+ *
+ * Emits the full signed manifest JSON to stdout. Used by the post-deploy
+ * CI job that publishes the manifest to GCS, reproducible by hand for
+ * local verification, and (MAT-2198) bundled by the SDK's own build into
+ * a standalone `dist/scripts/sign-trust-manifest.mjs` — a dependency-free
+ * ESM file shipped in the AWS release tarball as `bin/sign-trust-manifest.mjs`,
+ * runnable with plain `node` and no repo checkout / pnpm install, for
+ * clients who only ever have the downloaded release, not the source tree.
+ *
+ * It signs the SAME bytes the SDK verifies — it imports
+ * `canonicalManifestBytes` from the SDK source, so the signer and the
+ * `fetchTrustedManifest` verifier can never drift.
+ *
+ * Usage (TDX):
+ *   MANIFEST_OPERATOR_SIGNING_KEY=<hex-32-byte-privkey> \
+ *   tsx scripts/sign-trust-manifest.ts \
+ *     --cluster staging \
+ *     --version 1 \
+ *     --peer-ids '["12D3KooW...","12D3KooW...","12D3KooW..."]' \
+ *     --rtmr3 '["<base64-48-bytes>","<base64-48-bytes>"]' \
+ *     --rtmr1 '["<base64-48-bytes>","<base64-48-bytes>"]' \
+ *     --signed-at 2026-07-20T00:00:00Z
+ *
+ * Usage (all-SEV-SNP, e.g. an AWS-only fleet — MAT-2198):
+ *   MANIFEST_OPERATOR_SIGNING_KEY=<hex-32-byte-privkey> \
+ *   tsx scripts/sign-trust-manifest.ts \
+ *     --cluster my-aws-cluster \
+ *     --version 1 \
+ *     --peer-ids '["12D3KooW...","12D3KooW...","12D3KooW..."]' \
+ *     --sev-snp-measurements '["<base64-48-bytes>"]' \
+ *     --sev-snp-product Milan \
+ *     --signed-at 2026-07-20T00:00:00Z
+ *
+ * --rtmr3/--rtmr1 and --sev-snp-measurements/--sev-snp-product are each
+ * optional, but at least one vendor's pair must be non-empty — an anchor
+ * with nothing in either allow-list would trust any node. --rtmr3/--rtmr1
+ * describe the same TDX fleet and must be omitted/empty together, never
+ * just one. --sev-snp-product is required only when --sev-snp-measurements
+ * is non-empty.
+ *
+ * The private key is read only from the MANIFEST_OPERATOR_SIGNING_KEY
+ * environment variable (never a CLI arg, so it does not land in process
+ * listings / CI logs).
+ */
+function arg(name) {
+    const i = process.argv.indexOf(`--${name}`);
+    if (i < 0)
+        return undefined;
+    // A flag with no following value (e.g. --sev-snp-measurements as the
+    // final argv token) is a malformed invocation, not "flag omitted" -
+    // process.argv[i + 1] would otherwise silently return undefined for
+    // both cases, and a caller like the optional --sev-snp-measurements
+    // parsing would convert that into an empty allow-list and sign an
+    // all-TDX manifest instead of failing loudly.
+    if (i + 1 >= process.argv.length) {
+        throw new Error(`--${name} was given but no value followed it`);
+    }
+    return process.argv[i + 1];
+}
+function requireArg(name) {
+    const v = arg(name);
+    if (v === undefined) {
+        throw new Error(`missing required --${name}`);
+    }
+    return v;
+}
+function parseJsonStringArray(name, raw) {
+    const parsed = JSON.parse(raw);
+    if (!Array.isArray(parsed) || !parsed.every((x) => typeof x === "string")) {
+        throw new Error(`--${name} must be a JSON array of strings`);
+    }
+    return parsed;
+}
+function main() {
+    const privHex = process.env.MANIFEST_OPERATOR_SIGNING_KEY;
+    if (!privHex) {
+        throw new Error("MANIFEST_OPERATOR_SIGNING_KEY (hex secp256k1 private key) is required");
+    }
+    const peerIds = parseJsonStringArray("peer-ids", requireArg("peer-ids"));
+    if (peerIds.length === 0) {
+        throw new Error("peer-ids must be non-empty");
+    }
+    // --rtmr3/--rtmr1 are optional, like --sev-snp-measurements: an
+    // all-SEV-SNP fleet (the normal shape for an AWS-only deployment,
+    // MAT-2198) has no TDX measurements at all. They still describe the
+    // same TDX fleet together, though, so one present without the other
+    // is a real usage error, not a valid partial state.
+    const rtmr3Raw = arg("rtmr3");
+    const rtmr3 = rtmr3Raw === undefined ? [] : parseJsonStringArray("rtmr3", rtmr3Raw);
+    const rtmr1Raw = arg("rtmr1");
+    const rtmr1 = rtmr1Raw === undefined ? [] : parseJsonStringArray("rtmr1", rtmr1Raw);
+    if ((rtmr3.length === 0) !== (rtmr1.length === 0)) {
+        throw new Error("--rtmr3 and --rtmr1 must both be omitted/empty or both be non-empty " +
+            "(they describe the same TDX fleet)");
+    }
+    const sevSnpMeasurementsRaw = arg("sev-snp-measurements");
+    const sevSnpMeasurements = sevSnpMeasurementsRaw === undefined
+        ? []
+        : parseJsonStringArray("sev-snp-measurements", sevSnpMeasurementsRaw);
+    const sevSnpProduct = arg("sev-snp-product");
+    if (sevSnpMeasurements.length > 0 && sevSnpProduct === undefined) {
+        throw new Error("--sev-snp-product is required when --sev-snp-measurements is non-empty");
+    }
+    if (rtmr3.length === 0 && sevSnpMeasurements.length === 0) {
+        throw new Error("at least one of --rtmr3/--rtmr1 (TDX) or --sev-snp-measurements " +
+            "(SEV-SNP) must be non-empty — an empty anchor would trust any node");
+    }
+    const payload = {
+        cluster: requireArg("cluster"),
+        version: Number(requireArg("version")),
+        peer_ids: peerIds,
+        rtmr3_allowlist: rtmr3,
+        rtmr1_allowlist: rtmr1,
+        ...(sevSnpMeasurements.length > 0
+            ? { sev_snp_measurement_allowlist: sevSnpMeasurements }
+            : {}),
+        ...(sevSnpProduct === undefined ? {} : { sev_snp_product: sevSnpProduct }),
+        signed_at: requireArg("signed-at"),
+    };
+    if (!Number.isInteger(payload.version) || payload.version < 1) {
+        throw new Error("--version must be a positive integer");
+    }
+    const digest = canonicalManifestBytes(payload);
+    const signature = secp256k1.sign(digest, hexToBytes$1(privHex.replace(/^0x/, "")));
+    const manifest = {
+        ...payload,
+        signature: bytesToHex$1(signature),
+    };
+    process.stdout.write(`${JSON.stringify(manifest, null, 2)}\n`);
+}
+main();