OPENSSL-PKEYUTL(1SSL)

📦
version 4.0.3 · 13 versions
, in ubuntu stonking

💖 openssl-pkeyutl(1SSL) — the OpenSSL manual, but make it make sense 💖

okay bestie so this is openssl-pkeyutl from the OpenSSL manual pages, and it's literally the asymmetric key command, and we're about to get into it ✨

🌸 NAME 🌸

openssl-pkeyutl — asymmetric key command 👑

✨ SYNOPSIS ✨

openssl pkeyutl [-help] [-in file] [-rawin] [-digest algorithm] [-out file]
               [-secret file] [-sigfile file] [-inkey filename|uri]
               [-keyform DER|PEM|P12] [-passin arg] [-pubin] [-certin] [-rev]
               [-sign] [-verify] [-verifyrecover] [-encrypt] [-decrypt]
               [-derive] [-peerkey file] [-peerform DER|PEM|P12] [-encap]
               [-decap] [-kdf algorithm] [-kdflen length] [-kemop mode]
               [-pkeyopt opt:value] [-pkeyopt_passin opt[:passarg]] [-hexdump]
               [-asn1parse] [-rand files] [-writerand file] [-provider name]
               [-provider-path path] [-provparam [name:]key=value]
               [-propquery propq] [-config configfile]

🦋 DESCRIPTION 🦋

So basically this command does low-level operations on asymmetric keys — public or private — using any algorithm OpenSSL supports. she's flexible like that 💅

And here's the lil sneaky default you need to know: if you don't say what you wanna do, the signing operation is assumed (that's the -sign option, iykyk). so no vibe check needed, she'll just sign. 😭 iconic honestly.

🎀 OPTIONS 🎀

  • -help — prints out a usage message. classic. the "wait what do I do" button 🫶
  • -in filename — this is the input file you're reading data from. if you don't pass it, she reads from standard input. 💖
  • -rawin — ok so this one says: the signature or verification input data is raw data, not hashed by any message digest algorithm. For everything except EdDSA, you can still name a digest algorithm with the -digest option. For signature algorithms like RSA, DSA and ECDSA, the default digest algorithm is SHA256. For SM2, it's SM3. 🍓

This option can only be used with -sign and -verify. For EdDSA (that's the Ed25519 and Ed448 algorithms) this option is implied since OpenSSL 3.5, and required in earlier versions.

Also important: the -digest option implies -rawin since OpenSSL 3.5. periodt.

  • -digest algorithm — this one can only be used with -sign and -verify too. It says which digest algorithm gets used to hash the input data before you sign it or verify it with the input key. You can skip this option if your signature algorithm doesn't need to preprocess the input through a pluggable hash function before signing (EdDSA is the vibe here, for instance). If you omit it but the signature algorithm does need one and you gave -rawin, a default value gets used (go look at -rawin for the deets 💅). And if this option is present, then -rawin is implied since OpenSSL 3.5, and required in earlier versions.

Heads up bestie: right now, HashEdDSA (the ph or "prehash" variant of EdDSA) is not supported, so you cannot use -digest with EdDSA. 🚫

  • -out filename — the output filename to write to. defaults to standard output. ✨
  • -secret filename — the shared-secret output filename, for when you're doing encapsulation via -encap or decapsulation via -decap. And listen: -encap also produces a separate (public) ciphertext output, which by default goes to standard output — but since it's binary non-text data, you'll typically also redirect it to a file picked with -out. 💜
  • -sigfile file — signature file. required and allowed only for -verify operations. that's the whole job 🎀
  • -inkey filename|uri — the input key. by default it should be a private key. 🔑
  • -keyform DER|PEM|P12 — the key format. unspecified by default. see openssl-format-options(1) for the details. 🌸
  • -passin arg — the input key password source. for more info about the format of arg, see openssl-passphrase-options(1). 🫶
  • -pubin — by default a private key is read from the key input. with this option, a public key is read instead. and if the input has no public key but does have a private key, its public part gets used. slay, efficient 💅
  • -certin — the input is a certificate containing a public key. 📜✨
  • -rev — reverses the order of the input buffer. this is useful for some libraries (like CryptoAPI) that represent the buffer in little-endian format. cannot be used with -rawin — those two do NOT vibe together. 💀
  • -sign — sign the input data and output the signed result. this needs a private key. Using a message digest operation alongside this is recommended when applicable — check the -rawin and -digest options for the deets. Otherwise, the input data you gave with -in is assumed to already be a digest</italic>, but that may then require an additional -pkeyopt "digest:"md in some cases (for example, RSA with the default PKCS#1 padding mode). And even for other algorithms like ECDSA, where that extra -pkeyopt option doesn't affect the signature output, it's still recommended — because it lets you check that the input length is consistent with the digest you actually intended. main character attention to detail 👑🔥
  • -verify — verify the input data against the signature given with the -sigfile option, and tell you if verification succeeded or failed. The input data you gave with -in is assumed to be a hash value unless -rawin is specified or implied. With raw data, when a digest algorithm is applicable — even though it might get inferred from the signature or take a default value — you should also specify it. no guessing games 🎀
  • -verifyrecover — verify the given signature and output the recovered data (the signature payload). For example, in the case of RSA PKCS#1, the recovered data is the EMSA-PKCS-v1_5 DER encoding of the digest algorithm OID and value, as specified in RFC8017 Section 9.2 (https://datatracker.ietf.org/doc/html/rfc8017#section-9.2). 🦋

Note: here the input you gave with -in is not a signature input (like with -sign and -verify) — it's a signature output value, typically produced using -sign. yes, it's a whole switcheroo, keep up bestie 💅

This option is available only for use with RSA keys.

  • -encrypt — encrypt the input data using a public key. 🔐
  • -decrypt — decrypt the input data using a private key. 🔓
  • -derive — derive a shared secret using your own private (EC)DH key and the peer key. 🤝
  • -peerkey file — file containing the peer public or private (EC)DH key to use with the key derivation (agreement) operation. Its type must match the type of your own private key given with -inkey. matching sets only 🫶
  • -peerform DER|PEM|P12 — the peer key format. unspecified by default. see openssl-format-options(1) for the details. ✨
  • -encap — use a Key Encapsulation Mechanism (KEM) to encapsulate a shared-secret to a peer's public key. The encapsulated result (or ciphertext — non-text binary data) is written to standard output by default, or else to the file specified with -out. You must also provide -secret to specify the output file for the derived shared-secret value generated in the encapsulation process. Encapsulation is supported with a bunch of public key algorithms — currently: ML-KEM, X25519, X448, and EC. The ECX and EC algorithms use the RFC9180 (https://www.rfc-editor.org/rfc/rfc9180) DHKEM construction. Encapsulation is also supported with RSA keys via the RSASVE construction. 🎁✨

At the API level, encapsulation and decapsulation are also supported for a few hybrid ECDHE (no DHKEM) plus ML-KEM algorithms — but those are intended primarily for use with TLS and should not be used standalone. There are anyway no standard public and private key formats for the hybrid algorithms, so you literally can't provide the required key material. it's giving "don't try it" 💀

  • -decap — decode an encapsulated secret, using a -private key, to derive the same shared-secret as the one you'd get when the secret was encapsulated to the corresponding public key. The encapsulated secret is by default read from standard input, or else from the file specified with -in. The derived shared-secret is written to the file specified with -secret, which you must also provide. Decapsulation is supported with a bunch of public key algorithms — currently: ML-KEM, X25519, X448, and EC. The ECX and EC algorithms use the RFC9180 (https://www.rfc-editor.org/rfc/rfc9180) DHKEM construction. Decapsulation is also supported with RSA keys via the RSASVE construction. 🎁🔓
  • -kemop mode — this one hangs out with the -encap/-decap commands and specifies the KEM mode specific to the key algorithm, for when there's no default way to encapsulate and decapsulate shared secrets with the chosen key type. All the supported algorithms presently support only their default mode, and this option — though available — is not required. so like, she's there, she's cute, you don't need her 💅
  • -kdf algorithm — use key derivation function algorithm. The supported algorithms are at present TLS1-PRF and HKDF. Note: additional parameters and the KDF output length will normally have to be set for this to work. See EVP_PKEY_CTX_set_hkdf_md(3) and EVP_PKEY_CTX_set_tls1_prf_md(3) for the supported string parameters of each algorithm. 🧚
  • -kdflen length — set the output length for the KDF. 📏
  • -pkeyopt opt:value — public key options specified as opt:value. See NOTES below for more details. 👀
  • -pkeyopt_passin opt[:passarg] — lets you read a public key option opt from stdin or a password source. If you specify only opt, the user gets prompted to enter a password on stdin. Alternatively you can specify passarg, which can be any value supported by openssl-passphrase-options(1). 🔑
  • -hexdump — hex dump the output data. pretty colors of the binary kind 🌈
  • -asn1parse — parse the ASN.1 output data to check its DER encoding and print any errors. When combined with the -verifyrecover option, this may be useful in case an ASN.1 DER-encoded structure had been signed directly (without hashing it) and when checking a signature in PKCS#1 v1.5 format, which has a DER encoding. 🦋
  • -rand files, -writerand file — see "Random State Options" in openssl(1) for details. ✨
  • -provider name
  • -provider-path path
  • -provparam [name:]key=value
  • -propquery propq — see "Provider Options" in openssl(1), provider(7), and property(7). 🫶
  • -config configfile — see "Configuration Option" in openssl(1). 💖

more coming bestie, don't go anywhere 💅

🧠 NOTES 🧠

okay so real talk bestie — the operations and options that actually work depend entirely on the key algorithm and whatever implementation is backing it. The OpenSSL operations and options are laid out below, so read the section for your key type, okay? 💅

Unless it says otherwise somewhere, the -pkeyopt option supports the "digest:"alg argument for all public-key types — and that's what tells it which digest is in use for the signing and verification ops. The value alg should be a digest name like you'd use in the EVP_get_digestbyname() function, e.g. sha256. Important vibe check: this value does not hash your input data. It is used (by some algorithms) for sanity-checking the lengths of data passed in and for creating the structures that make up the signature (e.g. DigestInfo in RSASSA PKCS#1 v1.5 signatures).

So if the value of the -pkeyopt option "digest" argument is sha256, then your signature or verification input should be the 32 bytes long binary value of the SHA256 hash function output. Like, you hash it yourself first, bestie, pkeyutl is not gonna do it for you.

Unless -rawin is used or implied, this command does not hash the input data — it just takes your data straight into the signature algorithm, raw, no filter. Depending on the key type, signature type, and mode of padding, the maximum sensible lengths of input data differ. With RSA the signed data cannot be longer than the key modulus. In case of ECDSA and DSA the data should not be longer than the field size, otherwise it will be silently truncated to the field size (yes, silently, the audacity 😭). In any event the input size must not be larger than the largest supported digest output size EVP_MAX_MD_SIZE, which currently is 64 bytes.

🔐 RSA ALGORITHM 🔐

The RSA algorithm generally supports the encrypt, decrypt, sign, verify and verifyrecover operations. However, some padding modes only vibe with a subset of these. The following additional pkeyopt values are supported:

  • rsa_padding_mode:mode — sets the RSA padding mode. Acceptable values for mode are pkcs1 for PKCS#1 padding, none for no padding, oaep for OAEP mode, x931 for X9.31 mode and pss for PSS.

In PKCS#1 padding, if the message digest is not set, then the supplied data is signed or verified directly instead of using a DigestInfo structure. If a digest is set, then the DigestInfo structure is used and its length must correspond to the digest type.

Note, for pkcs1 padding, as a protection against the Bleichenbacher attack, the decryption will not fail in case of padding check failures. Use none and manual inspection of the decrypted message to verify if the decrypted value has correct PKCS#1 v1.5 padding. 👀

For oaep mode only encryption and decryption is supported.

For x931 if the digest type is set it is used to format the block data, otherwise the first byte is used to specify the X9.31 digest ID. Sign, verify and verifyrecover can be performed in this mode.

For pss mode only sign and verify are supported and the digest type must be specified.

  • rsa_pss_saltlen:len — for pss mode only, this option specifies the salt length. Three special values are supported: digest sets the salt length to the digest length, max sets the salt length to the maximum permissible value. When verifying, auto causes the salt length to be automatically determined based on the PSS block structure. ✨
  • rsa_mgf1_md:digest — for PSS and OAEP padding, sets the MGF1 digest. If the MGF1 digest is not explicitly set in PSS mode, then the signing digest is used.
  • rsa_oaep_md:digest — sets the digest used for the OAEP hash function. If not explicitly set, then SHA1 is used.
  • rsa_pkcs1_implicit_rejection:flag — disables (when set to 0) or enables (when set to 1) the use of implicit rejection with PKCS#1 v1.5 decryption. When enabled (the default), as a protection against Bleichenbacher attack, the library will generate a deterministic random plaintext that it will return to the caller in case of padding check failure. When disabled, it's the callers' responsibility to handle the returned errors in a side-channel free manner. no pressure or anything 💀

🌈 RSA-PSS ALGORITHM 🌈

The RSA-PSS algorithm is a restricted version of the RSA algorithm which only supports the sign and verify operations with PSS padding. The following additional -pkeyopt values are supported:

  • rsa_padding_mode:mode, rsa_pss_saltlen:len, rsa_mgf1_md:digest — these have the same meaning as the RSA algorithm with some additional restrictions. The padding mode can only be set to pss which is the default value.

If the key has parameter restrictions then the digest, MGF1 digest and salt length are set to the values specified in the parameters. The digest and MG cannot be changed and the salt length cannot be set to a value less than the minimum restriction. 🔒

🍓 DSA ALGORITHM 🍓

The DSA algorithm supports signing and verification operations only. Currently there are no additional -pkeyopt options other than digest. The SHA256 digest is assumed by default.

🦋 DH ALGORITHM 🦋

The DH algorithm only supports the derivation operation and no additional -pkeyopt options. that's it, that's the section 🫶

🎀 EC ALGORITHM 🎀

The EC algorithm supports sign, verify and derive operations. The sign and verify operations use ECDSA and derive uses ECDH. SHA256 is assumed by default for the -pkeyopt digest option.

💫 X25519 AND X448 ALGORITHMS 💫

The X25519 and X448 algorithms support key derivation only. Currently there are no additional options.

🔮 SLH-DSA ALGORITHMS 🔮

The SLH-DSA algorithms (SLH-DSA-SHA2-128s, SLH-DSA-SHA2-128f, SLH-DSA-SHA2-192s, SLH-DSA-SHA2-192f, SLH-DSA-SHA2-256s, SLH-DSA-SHA2-256f) are post-quantum signature algorithms. When using SLH-DSA with pkeyutl, the following options are available:

  • -sign — sign the input data using an SLH-DSA private key. For example:
$ openssl pkeyutl -sign -in file.txt -inkey slhdsa.pem -out sig
  • -verify — verify the signature using an SLH-DSA public key. For example:
$ openssl pkeyutl -verify -in file.txt -inkey slhdsa.pem -sigfile sig

See EVP_PKEY-SLH-DSA(7) and EVP_SIGNATURE-SLH-DSA(7) for additional details about the SLH-DSA algorithm and its implementation.

👑 ML-DSA-44, ML-DSA-65 AND ML-DSA-87 ALGORITHMS 👑

The ML-DSA algorithms are post-quantum signature algorithms that support signing and verification of "raw" messages. No preliminary hashing is performed. When using ML-DSA with pkeyutl, the following options are available:

  • -sign — sign the input data using an ML-DSA private key. For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig
  • -verify — verify the signature using an ML-DSA public key. For example:
$ openssl pkeyutl -verify -in file.txt -inkey mldsa65.pem -sigfile sig
  • -pkeyopt opt:value — additional options for ML-DSA signing and verification:
    • message-encoding:value — specifies the message encoding mode used for signing. This controls how the input message is processed before signing. Valid values are described in EVP_SIGNATURE-ML-DSA(7). For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt message-encoding:1
  • test-entropy:value — specifies a test entropy value for deterministic signing. For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt test-entropy:abcdefghijklmnopqrstuvwxyz012345
  • hextest-entropy:value — specifies a test entropy value in hex format. For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt hextest-entropy:000102030405060708090a0b0c0d0e0f101112131415161718191a1b1c1d1e1f
  • deterministic:value — enables deterministic signing. For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt deterministic:1
  • mu:value — specifies the mu parameter. For example:
$ echo -n "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef" >file.txt
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt mu:1
  • context-string:string — specifies a context string for both signing and verification operations. The context string must be the same for verification to succeed. For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt context-string:mycontext
$ openssl pkeyutl -verify -in file.txt -inkey mldsa65.pem -sigfile sig -pkeyopt context-string:mycontext
  • hexcontext-string:string — specifies a context string in hex format, allowing binary control values. For example:
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt hexcontext-string:6d79636f6e74657874

The signing operation supports a deterministic:bool option, with bool set to 1 if a deterministic signature is to be generated with a fixed all zero random input. By default, or if the bool is 0, a random entropy value is used. A deterministic result can also be obtained by specifying an explicit entropy value via the hextest-entropy:value parameter. Deterministic ML-DSA signing should only be used in tests. ⚠️

See EVP_SIGNATURE-ML-DSA(7) for additional details about the ML-DSA algorithms and their implementation.

🧚 ML-KEM-512, ML-KEM-768 AND ML-KEM-1024 ALGORITHMS 🧚

The ML-KEM algorithms support encapsulation and decapsulation only. The encapsulation operation supports a hexikme:entropy option, with entropy the 64 hexadecimal digit encoding of a 32-byte value. This should only be used in tests — known or leaked values of the option may compromise the generated shared secret. 😳

See EVP_KEM-ML-KEM(7) for additional detail.

⭐ ED25519 AND ED448 ALGORITHMS ⭐

These algorithms only support signing and verifying. OpenSSL only implements the "pure" variants of these algorithms, so raw data can be passed directly to them without hashing them first. OpenSSL only supports "oneshot" operation with these algorithms. This means that the entire file to be signed/verified must be read into memory before processing it. Signing or Verifying very large files should be avoided. Additionally the size of the file must be known for this to work. If the size of the file cannot be determined (for example if the input is stdin) then the sign or verify operation will fail. 😭

✨ SM2 ✨

okay so SM2 is the one that does it all — sign, verify, encrypt and decrypt operations, the whole vibe. 💅

but bestie, for the sign and verify operations specifically? SM2 needs a Distinguishing ID string to be passed in. it won't do it without one. so here's the -pkeyopt value that's supported:

  • distid:string — sets the ID string used in SM2 sign or verify operations. 🫶 and listen: when you're verifying an SM2 signature, the ID string must be the exact same one used when signing the data. Otherwise the verification will fail. no cap, it will fail.
  • hexdistid:hex_string — same deal, sets the ID string used in SM2 sign or verify operations. While verifying an SM2 signature, the ID string must be the same one used when signing the data. Otherwise the verification will fail. but the ID string you provide with this option should be a valid hexadecimal value. ✨

🎀 EXAMPLES 🎀

Sign some data using a private key:

openssl pkeyutl -sign -in file -inkey key.pem -out sig

Recover the signed data (e.g. if an RSA key is used):

openssl pkeyutl -verifyrecover -in sig -inkey key.pem

Verify the signature (e.g. a DSA key):

openssl pkeyutl -verify -in file -sigfile sig -inkey key.pem

Sign data using a message digest value (this is currently only valid for RSA):

openssl pkeyutl -sign -in file -inkey key.pem -out sig -pkeyopt digest:sha256

Derive a shared secret value:

openssl pkeyutl -derive -inkey key.pem -peerkey pubkey.pem -out secret

Hexdump 48 bytes of TLS1 PRF using digest SHA256 and shared secret and seed consisting of the single byte 0xFF:

openssl pkeyutl -kdf TLS1-PRF -kdflen 48 -pkeyopt md:SHA256 \
   -pkeyopt hexsecret:ff -pkeyopt hexseed:ff -hexdump

Derive a key using scrypt where the password is read from command line:

openssl pkeyutl -kdf scrypt -kdflen 16 -pkeyopt_passin pass \
   -pkeyopt hexsalt:aabbcc -pkeyopt N:16384 -pkeyopt r:8 -pkeyopt p:1

Derive using the same algorithm, but read key from environment variable MYPASS:

openssl pkeyutl -kdf scrypt -kdflen 16 -pkeyopt_passin pass:env:MYPASS \
   -pkeyopt hexsalt:aabbcc -pkeyopt N:16384 -pkeyopt r:8 -pkeyopt p:1

Sign some data using an SM2(7) private key and a specific ID:

openssl pkeyutl -sign -in file -inkey sm2.key -out sig -rawin -digest sm3 \
   -pkeyopt distid:someid

Verify some data using an SM2(7) certificate and a specific ID:

openssl pkeyutl -verify -certin -in file -inkey sm2.cert -sigfile sig \
   -rawin -digest sm3 -pkeyopt distid:someid

Decrypt some data using a private key with OAEP padding using SHA256:

openssl pkeyutl -decrypt -in file -inkey key.pem -out secret \
   -pkeyopt rsa_padding_mode:oaep -pkeyopt rsa_oaep_md:sha256

Create an ML-DSA key pair and sign data with a specific context string:

$ openssl genpkey -algorithm ML-DSA-65 -out mldsa65.pem
$ openssl pkeyutl -sign -in file.txt -inkey mldsa65.pem -out sig -pkeyopt context-string:example

Verify a signature using ML-DSA with the same context string:

$ openssl pkeyutl -verify -in file.txt -inkey mldsa65.pem -sigfile sig -pkeyopt context-string:example

Generate an ML-KEM key pair and use it for encapsulation:

$ openssl genpkey -algorithm ML-KEM-768 -out mlkem768.pem
$ openssl pkey -in mlkem768.pem -pubout -out mlkem768_pub.pem
$ openssl pkeyutl -encap -inkey mlkem768_pub.pem -pubin -out ciphertext -secret shared_secret.bin

Decapsulate a shared secret using an ML-KEM private key:

$ openssl pkeyutl -decap -inkey mlkem768.pem -in ciphertext -secret decapsulated_secret.bin

Create an SLH-DSA key pair and sign data:

$ openssl genpkey -algorithm SLH-DSA-SHA2-128s -out slh-dsa.pem
$ openssl pkeyutl -sign -in file.txt -inkey slh-dsa.pem -out sig

Verify a signature using SLH-DSA:

$ openssl pkeyutl -verify -in file.txt -inkey slh-dsa.pem -sigfile sig

🌈 SEE ALSO 🌈

openssl(1), openssl-genpkey(1), openssl-pkey(1), openssl-rsautl(1) openssl-dgst(1), openssl-rsa(1), openssl-genrsa(1), openssl-kdf(1) EVP_PKEY_CTX_set_hkdf_md(3), EVP_PKEY_CTX_set_tls1_prf_md(3),

📜 HISTORY 📜

the lore!! 🫶

Since OpenSSL 3.5, the -digest option implies -rawin. and the -rawin option is no longer required when signing or verifying with a key type that does not support a prehash digest, such as Ed25519, Ed448, ML-DSA, or SLH-DSA. for these key types, -digest is not supported. so don't try it. 🙈

Also since OpenSSL 3.5, the -kemop option is no longer required for any of the supported algorithms — the only supported mode is now the default. iconic behaviour tbh. ✨

The -engine option was removed in OpenSSL 4.0. rip bozo, it's gone, do not pass go. 💀

Copyright 2006-2026 The OpenSSL Project Authors. All Rights Reserved.

Licensed under the Apache License 2.0 (the "License"). You may not use this file except in compliance with the License. You can obtain a copy in the file LICENSE in the source distribution or at <https://www.openssl.org/source/license.html>.

and that's a wrap bestie 🎀💖 this whole page is OpenSSL 4.0.3, dated 2026-09-29, and it lives under the name OPENSSL-PKEYUTL(1SSL). go forth and pkeyutl responsibly, mwah 💅✨

📦 where this page comes from