Off-the-Record Messaging Protocol version 3

This document describes version 3 of the Off-the-Record Messaging protocol. The main changes over version 2 include:

Very high level overview

OTR assumes a network model which provides in-order delivery of messages, but that some messages may not get delivered at all (for example, if the user disconnects). There may be an active attacker, who is allowed to perform a Denial of Service attack, but not to learn the contents of messages.

  1. Alice signals to Bob that she would like (using an OTR Query Message) or is willing (using a whitespace-tagged plaintext message) to use OTR to communicate. Either mechanism should convey the version(s) of OTR that Alice is willing to use.
  2. Bob initiates the authenticated key exchange (AKE) with Alice. Versions 2 and 3 of OTR use a variant of the SIGMA protocol as its AKE.
  3. Alice and Bob exchange Data Messages to send information to each other.

High level overview

Requesting an OTR conversation

There are two ways Alice can inform Bob that she is willing to use the OTR protocol to speak with him: by sending him the OTR Query Message, or by including a special "tag" consisting of whitespace characters in one of her messages to him. Each method also includes a way for Alice to communicate to Bob which versions of the OTR protocol she is willing to speak with him.

The semantics of the OTR Query Message are that Alice is requesting that Bob start an OTR conversation with her (if, of course, he is willing and able to do so). On the other hand, the semantics of the whitespace tag are that Alice is merely indicating to Bob that she is willing and able to have an OTR conversation with him. If Bob has a policy of "only use OTR when it's explicitly requested", for example, then he would start an OTR conversation upon receiving an OTR Query Message, but would not upon receiving the whitespace tag.

Authenticated Key Exchange (AKE)

This section outlines the version of the SIGMA protocol used as the AKE. All exponentiations are done modulo a particular 1536-bit prime, and g is a generator of that group, as indicated in the detailed description below. Alice and Bob's long-term authentication public keys are pubA and pubB, respectively.

The general idea is that Alice and Bob do an unauthenticated Diffie-Hellman (D-H) key exchange to set up an encrypted channel, and then do mutual authentication inside that channel.

Bob will be initiating the AKE with Alice.

Exchanging data

This section outlines the method used to protect data being exchanged between Alice and Bob. As above, all exponentiations are done modulo a particular 1536-bit prime, and g is a generator of that group, as indicated in the detailed description below.

Suppose Alice has a message (msg) to send to Bob.

Socialist Millionaires' Protocol (SMP)

While data messages are being exchanged, either Alice or Bob may run SMP to detect impersonation or man-in-the-middle attacks. As above, all exponentiations are done modulo a particular 1536-bit prime, and g1 is a generator of that group. All sent values include zero-knowledge proofs that they were generated according to this protocol, as indicated in the detailed description below.

Suppose Alice and Bob have secret information x and y respectively, and they wish to know whether x = y. The Socialist Millionaires' Protocol allows them to compare x and y without revealing any other information than the value of (x == y). For OTR, the secrets contain information about both parties' long-term authentication public keys, as well as information entered by the users themselves. If x = y, this means that Alice and Bob entered the same secret information, and so must be the same entities who established that secret to begin with.

Assuming that Alice begins the exchange:

Details of the protocol

Unencoded messages

This section describes the messages in the OTR protocol that are not base-64 encoded binary.

OTR Query Messages

If Alice wishes to communicate to Bob that she would like to use OTR, she sends a message containing the string "?OTR" followed by an indication of what versions of OTR she is willing to use with Bob. The version string is constructed as follows:

For example:

"?OTR?"
Version 1 only
"?OTRv2?"
Version 2 only
"?OTRv23?"
Versions 2 and 3
"?OTR?v2?"
Versions 1 and 2
"?OTRv24x?"
Version 2, and hypothetical future versions identified by "4" and "x"
"?OTR?v24x?"
Versions 1, 2, and hypothetical future versions identified by "4" and "x"
"?OTR?v?"
Also version 1 only
"?OTRv?"
A bizarre claim that Alice would like to start an OTR conversation, but is unwilling to speak any version of the protocol

These strings may be hidden from the user (for example, in an attribute of an HTML tag), and/or may be accompanied by an explanitory message ("Alice has requested an Off-the-Record private conversation."). If Bob is willing to use OTR with Alice (with a protocol version that Alice has offered), he should start the AKE.

Tagged plaintext messages

If Alice wishes to communicate to Bob that she is willing to use OTR, she can attach a special whitespace tag to any plaintext message she sends him. This tag may occur anywhere in the message, and may be hidden from the user (as in the Query Messages, above).

The tag consists of the following 16 bytes, followed by one or more sets of 8 bytes indicating the version of OTR Alice is willing to use:

If Bob is willing to use OTR with Alice (with a protocol version that Alice has offered), he should start the AKE. On the other hand, if Alice receives a plaintext message from Bob (rather than an initiation of the AKE), she should stop sending him the whitespace tag.

OTR Error Messages

Any message containing the string "?OTR Error:" is an OTR Error Message. The following part of the message should contain human-readable details of the error.

Encoded messages

This section describes the byte-level format of the base-64 encoded binary OTR messages. The binary form of each of the messages is described below. To transmit one of these messages, construct the ASCII string consisting of the five bytes "?OTR:", followed by the base-64 encoding of the binary form of the message, followed by the byte ".".

For the Diffie-Hellman group computations, the group is the one defined in RFC 3526 with 1536-bit modulus (hex, big-endian):

FFFFFFFF FFFFFFFF C90FDAA2 2168C234 C4C6628B 80DC1CD1
29024E08 8A67CC74 020BBEA6 3B139B22 514A0879 8E3404DD
EF9519B3 CD3A431B 302B0A6D F25F1437 4FE1356D 6D51C245
E485B576 625E7EC6 F44C42E9 A637ED6B 0BFF5CB6 F406B7ED
EE386BFB 5A899FA5 AE9F2411 7C4B1FE6 49286651 ECE45B3D
C2007CB8 A163BF05 98DA4836 1C55D39A 69163FA8 FD24CF5F
83655D23 DCA3AD96 1C62F356 208552BB 9ED52907 7096966D
670C354E 4ABC9804 F1746C08 CA237327 FFFFFFFF FFFFFFFF

and a generator (g) of 2. Note that this means that whenever you see a Diffie-Hellman exponentiation in this document, it always means that the exponentiation is done modulo the above 1536-bit number.

Data types

Bytes (BYTE):
1 byte unsigned value
Shorts (SHORT):
2 byte unsigned value, big-endian
Ints (INT):
4 byte unsigned value, big-endian
Multi-precision integers (MPI):
4 byte unsigned len, big-endian
len byte unsigned value, big-endian
(MPIs must use the minimum-length encoding; i.e. no leading 0x00 bytes. This is important when calculating public key fingerprints.)
Opaque variable-length data (DATA):
4 byte unsigned len, big-endian
len byte data
Initial CTR-mode counter value (CTR):
8 bytes data
Message Authentication Code (MAC):
20 bytes MAC data

Public keys, signatures, and fingerprints

OTR users have long-lived public keys that they use for authentication (but not encryption). The current version of the OTR protocol only supports DSA public keys, but there is a key type marker for future extensibility.

OTR public authentication DSA key (PUBKEY):
Pubkey type (SHORT) p (MPI)
q (MPI)
g (MPI)
y (MPI)

OTR public keys are used to generate signatures; different types of keys produce signatures in different formats. The format for a signature made by a DSA public key is as follows:

DSA signature (SIG):
(len is the length of the DSA public parameter q, which in current implementations must be 20 bytes, or 160 bits)
len byte unsigned r, big-endian
len byte unsigned s, big-endian

OTR public keys have fingerprints, which are hex strings that serve as identifiers for the public key. The fingerprint is calculated by taking the SHA-1 hash of the byte-level representation of the public key. However, there is an exception for backwards compatibility: if the pubkey type is 0x0000, those two leading 0x00 bytes are omitted from the data to be hashed. The encoding assures that, assuming the hash function itself has no useful collisions, and DSA keys have length less than 524281 bits (500 times larger than most DSA keys), no two public keys will have the same fingerprint.

Instance Tags

Clients include instance tags in all OTR version 3 messages. Instance tags are 32-bit values that are intended to be persistent. If the same client is logged into the same account from multiple locations, the intention is that the client will have different instance tags at each location. As shown below, OTR version 3 messages (fragmented and unfragmented) include the source and destination instance tags. If a client receives a message that lists a destination instance tag different from its own, the client should discard the message.

The smallest valid instance tag is 0x00000100. It is appropriate to set the destination instance tag to '0' when an actual destination instance tag is not known at the time the message is prepared. If a client receives a message with the sender instance tag set to less than 0x00000100, it should discard the message. Similarly, if a client receives a message with the recipient instance tag set to greater than 0 but less than 0x00000100, it should discard the message.

This avoids an issue on IM networks that always relay all messages to all sessions of a client who is logged in multiple times. In this situation, OTR clients can attempt to establish an OTR session indefinitely if there are interleaving messages from each of the sessions.

D-H Commit Message

This is the first message of the AKE. Bob sends it to Alice to commit to a choice of D-H encryption key (but the key itself is not yet revealed). This allows the secure session id to be much shorter than in OTR version 1, while still preventing a man-in-the-middle attack on it.

Protocol version (SHORT)
The version number of this protocol is 0x0003.
Message type (BYTE)
The D-H Commit Message has type 0x02.
Sender Instance tag (INT)
The instance tag of the person sending this message.
Receiver Instance tag (INT)
The instance tag of the intended recipient. For a commit message this will often be 0, since the other party may not have identified their instance tag yet.
Encrypted gx (DATA)
Produce this field as follows:
Hashed gx (DATA)
This is the SHA256 hash of gxmpi.

D-H Key Message

This is the second message of the AKE. Alice sends it to Bob, and it simply consists of Alice's D-H encryption key.

Protocol version (SHORT)
The version number of this protocol is 0x0003.
Message type (BYTE)
The D-H Key Message has type 0x0a.
Sender Instance tag (INT)
The instance tag of the person sending this message.
Receiver Instance tag (INT)
The instance tag of the intended recipient.
gy (MPI)
Choose a random value y (at least 320 bits), and calculate gy.

Reveal Signature Message

This is the third message of the AKE. Bob sends it to Alice, revealing his D-H encryption key (and thus opening an encrypted channel), and also authenticating himself (and the parameters of the channel, preventing a man-in-the-middle attack on the channel itself) to Alice.

Protocol version (SHORT)
The version number of this protocol is 0x0003.
Message type (BYTE)
The Reveal Signature Message has type 0x11.
Sender Instance tag (INT)
The instance tag of the person sending this message.
Receiver Instance tag (INT)
The instance tag of the intended recipient.
Revealed key (DATA)
This is the value r picked earlier.
Encrypted signature (DATA)
This field is calculated as follows:
MAC'd signature (MAC)
This is the SHA256-HMAC-160 (that is, the first 160 bits of the SHA256-HMAC) of the encrypted signature field (including the four-byte length), using the key m2.

Signature Message

This is the final message of the AKE. Alice sends it to Bob, authenticating herself and the channel parameters to him.

Protocol version (SHORT)
The version number of this protocol is 0x0003.
Message type (BYTE)
The Signature Message has type 0x12.
Sender Instance tag (INT)
The instance tag of the person sending this message.
Receiver Instance tag (INT)
The instance tag of the intended recipient.
Encrypted signature (DATA)
This field is calculated as follows:
MAC'd signature (MAC)
This is the SHA256-HMAC-160 (that is, the first 160 bits of the SHA256-HMAC) of the encrypted signature field (including the four-byte length), using the key m2'.

Data Message

This message is used to transmit a private message to the correspondent. It is also used to reveal old MAC keys.

The plaintext message (either before encryption, or after decryption) consists of a human-readable message (encoded in UTF-8, optionally with HTML markup), optionally followed by:

Each TLV record is of the form:

Type (SHORT)
The type of this record. Records with unrecognized types should be ignored.
Length (SHORT)
The length of the following field
Value (len BYTEs) [where len is the value of the Length field]
Any pertinent data for the record type.

Some TLV examples:

\x00\x01\x00\x00
A TLV of type 1, containing no data
\x00\x00\x00\x05\x68\x65\x6c\x6c\x6f
A TLV of type 0, containing the value "hello"

The currently defined TLV record types are:

Type 0: Padding
The value may be an arbitrary amount of data, which should be ignored. This type can be used to disguise the length of the plaintext message.
Type 1: Disconnected
If the user requests to close the private connection, you may send a message (possibly with empty human-readable part) containing a record with this TLV type just before you discard the session keys, and transition to MSGSTATE_PLAINTEXT (see below). If you receive a TLV record of this type, you should transition to MSGSTATE_FINISHED (see below), and inform the user that his correspondent has closed his end of the private connection, and the user should do the same.
Type 2: SMP Message 1
The value represents an initiating message of the Socialist Millionaires' Protocol, described below.
Type 3: SMP Message 2
The value represents the second message in an instance of SMP.
Type 4: SMP Message 3
The value represents the third message in an instance of SMP.
Type 5: SMP Message 4
The value represents the final message in an instance of SMP.
Type 6: SMP Abort Message
If the user cancels SMP prematurely or encounters an error in the protocol and cannot continue, you may send a message (possibly with empty human-readable part) with this TLV type to instruct the other party's client to abort the protocol. The associated length should be zero and the associated value should be empty. If you receive a TLV of this type, you should change the SMP state to SMP_EXPECT1 (see below).
Type 7: SMP Message 1Q
Like a SMP Message 1, but whose value begins with a NUL-terminated user-specified question.
Type 8: Extra symmetric key
If you wish to use the extra symmetric key, compute it yourself as outlined in the section "Extra symmetric key", below. Then send this type 8 TLV to your buddy to indicate that you'd like to use the extra symmetric key for something. The value of the TLV begins with a 4-byte indication of what this symmetric key will be used for (file transfer, voice encryption, etc.). After that, the contents are use-specific (which file, etc.). There are no currently defined uses. Note that the value of the key itself is not placed into the TLV; your buddy will compute it on his/her own.

SMP Message TLVs (types 2-5) all carry data sharing the same general format:

MPI count (INT)
The number of MPIs contained in the remainder of the TLV.
MPI 1 (MPI)
The first MPI of the TLV, serialized into a byte array.
MPI 2 (MPI)
The second MPI of the TLV, serialized into a byte array.
etc.

There should be as many MPIs as declared in the MPI count field. For the exact MPIs passed for each SMP TLV, see the SMP state machine below.

A message with an empty human-readable part (the plaintext is of zero length, or starts with a NUL) is a "heartbeat" packet, and should not be displayed to the user. (But it's still useful to effect key rotations.)

Data Message format:

Protocol version (SHORT)
The version number of this protocol is 0x0003.
Message type (BYTE)
The Data Message has type 0x03.
Sender Instance tag (INT)
The instance tag of the person sending this message.
Receiver Instance tag (INT)
The instance tag of the intended recipient.
Flags (BYTE)
The bitwise-OR of the flags for this message. Usually you should set this to 0x00. The only currently defined flag is:
IGNORE_UNREADABLE (0x01)
If you receive a Data Message with this flag set, and you are unable to decrypt the message or verify the MAC (because, for example, you don't have the right keys), just ignore the message instead of producing some kind of error or notification to the user.
Sender keyid (INT)
Must be strictly greater than 0, and increment by 1 with each key change
Recipient keyid (INT)
Must therefore be strictly greater than 0, as the receiver has no key with id 0.
The sender and recipient keyids are those used to encrypt and MAC this message.
DH y (MPI)
The *next* [i.e. sender_keyid+1] public key for the sender
Top half of counter init (CTR)
This should monotonically increase (as a big-endian value) for each message sent with the same (sender keyid, recipient keyid) pair, and must not be all 0x00.
Encrypted message (DATA)
Using the appropriate encryption key (see below) derived from the sender's and recipient's DH public keys (with the keyids given in this message), perform AES128 counter-mode (CTR) encryption of the message. The initial counter is a 16-byte value whose first 8 bytes are the above "top half of counter init" value, and whose last 8 bytes are all 0x00. Note that counter mode does not change the length of the message, so no message padding needs to be done. If you *want* to do message padding (to disguise the length of your message), use the above TLV of type 0.
Authenticator (MAC)
The SHA1-HMAC, using the appropriate MAC key (see below) of everything from the Protocol version to the end of the encrypted message
Old MAC keys to be revealed (DATA)
See "Revealing MAC Keys", below.

Socialist Millionaires' Protocol (SMP)

The Socialist Millionaires' Protocol allows two parties with secret information x and y respectively to check whether (x==y) without revealing any additional information about the secrets. The protocol used by OTR is based on the work of Boudot, Schoenmakers and Traore (2001). A full justification for its use in OTR is made by Alexander and Goldberg, in a paper published in 2007. The following is a technical account of what is transmitted during the course of the protocol.

Secret information

The secret information x and y compared during this protocol contains not only information entered by the users, but also information unique to the conversation in which SMP takes place. Specifically, the format is:

Version (BYTE)
The version of SMP used. The version described here is 1.
Initiator fingerprint (20 BYTEs)
The fingerprint that the party initiating SMP is using in the current conversation.
Responder fingerprint (20 BYTEs)
The fingerprint that the party that did not initiate SMP is using in the current conversation.
Secure Session ID
The ssid described below.
User-specified secret
The input string given by the user at runtime.

Then the SHA256 hash of the above is taken, and the digest becomes the actual secret (x or y) to be used in SMP. The additional fields insure that not only do both parties know the same secret input string, but no man-in-the-middle is capable of reading their communication either.

The SMP state machine

Whenever the OTR message state machine has MSGSTATE_ENCRYPTED set (see below), the SMP state machine may progress. If at any point MSGSTATE_ENCRYPTED becomes unset, SMP must abandon its state and return to its initial setup. The SMP state consists of one main variable, as well as information from the partial computations at each protocol step.

Expected Message

This main state variable for SMP controls what SMP-specific TLVs will be accepted. This variable has no effect on type 0 or type 1 TLVs, which are always allowed. smpstate can take one of four values:

SMPSTATE_EXPECT1
This state indicates that only type 2 (SMP message 1) and type 7 (SMP message 1Q) TLVs should be accepted. This is the default state when SMP has not yet begun. This state is also reached whenever an error occurs or SMP is aborted, and the protocol must be restarted from the beginning.
SMPSTATE_EXPECT2
This state indicates that only type 3 TLVs (SMP message 2) should be accepted.
SMPSTATE_EXPECT3
This state indicates that only type 4 TLVs (SMP message 3) should be accepted.
SMPSTATE_EXPECT4
This state indicates that only type 5 TLVs (SMP message 4) should be accepted.

State Transitions

There are 7 actions that an OTR client must handle:

The following sections outline what is to be done in each case. They all assume that MSGSTATE_ENCRYPTED is set. For simplicity, they also assume that Alice has begun SMP, and Bob is responding to her.

SMP Hash function

In the following actions, there are many places where a SHA256 hash of an integer followed by one or two MPIs is taken. The input to this hash function is:

Version (BYTE)
This distinguishes calls to the hash function at different points in the protocol, to prevent Alice from replaying Bob's zero knowledge proofs or vice versa.
First MPI (MPI)
The first MPI given as input, serialized in the usual way.
Second MPI (MPI)
The second MPI given as input, if present, serialized in the usual way. If only one MPI is given as input, this field is simply omitted.

Receiving a type 2 TLV (SMP message 1)

SMP message 1 is sent by Alice to begin a DH exchange to determine two new generators, g2 and g3. It contains the following mpi values:

g2a
Alice's half of the DH exchange to determine g2.
c2, D2
A zero-knowledge proof that Alice knows the exponent associated with her transmitted value g2a.
g3a
Alice's half of the DH exchange to determine g3.
c3, D3
A zero-knowledge proof that Alice knows the exponent associated with her transmitted value g3a.

A type 7 (SMP Message 1Q) TLV is the same as the above, but is preceded by a user-specified question, which is associated with the user-specified portion of the secret.

When Bob receives this TLV he should do:

If smpstate is not SMPSTATE_EXPECT1:
Set smpstate to SMPSTATE_EXPECT1 and send a type 6 TLV (SMP abort) to Alice.
If smpstate is SMPSTATE_EXPECT1:
Verify Alice's zero-knowledge proofs for g2a and g3a:
  1. Check that both g2a and g3a are >= 2 and <= modulus-2.
  2. Check that c2 = SHA256(1, g1D2 g2ac2).
  3. Check that c3 = SHA256(2, g1D3 g3ac3).
Create a type 3 TLV (SMP message 2) and send it to Alice:
  1. Determine Bob's secret input y, which is to be compared to Alice's secret x.
  2. Pick random exponents b2 and b3. These will used during the DH exchange to pick generators.
  3. Pick random exponents r2, r3, r4, r5 and r6. These will be used to add a blinding factor to the final results, and to generate zero-knowledge proofs that this message was created honestly.
  4. Compute g2b = g1b2 and g3b = g1b3
  5. Generate a zero-knowledge proof that the exponent b2 is known by setting c2 = SHA256(3, g1r2) and D2 = r2 - b2 c2 mod q. In the zero-knowledge proofs the D values are calculated modulo q = (p - 1) / 2, where p is the same 1536-bit prime as elsewhere. The random exponents are 1536-bit numbers.
  6. Generate a zero-knowledge proof that the exponent b3 is known by setting c3 = SHA256(4, g1r3) and D3 = r3 - b3 c3 mod q.
  7. Compute g2 = g2ab2 and g3 = g3ab3
  8. Compute Pb = g3r4 and Qb = g1r4 g2y
  9. Generate a zero-knowledge proof that Pb and Qb were created according to the protocol by setting cP = SHA256(5, g3r5, g1r5 g2r6), D5 = r5 - r4 cP mod q and D6 = r6 - y cP mod q.
  10. Store the values of g3a, g2, g3, b3, Pb and Qb for use later in the protocol.
  11. Send Alice a type 3 TLV (SMP message 2) containing g2b, c2, D2, g3b, c3, D3, Pb, Qb, cP, D5 and D6, in that order.
Set smpstate to SMPSTATE_EXPECT3.

Receiving a type 3 TLV (SMP message 2)

SMP message 2 is sent by Bob to complete the DH exchange to determine the new generators, g2 and g3. It also begins the construction of the values used in the final comparison of the protocol. It contains the following mpi values:

g2b
Bob's half of the DH exchange to determine g2.
c2, D2
A zero-knowledge proof that Bob knows the exponent associated with his transmitted value g2b.
g3b
Bob's half of the DH exchange to determine g3.
c3, D3
A zero-knowledge proof that Bob knows the exponent associated with his transmitted value g3b.
Pb, Qb
These values are used in the final comparison to determine if Alice and Bob share the same secret.
cP, D5, D6
A zero-knowledge proof that Pb and Qb were created according to the protcol given above.

When Alice receives this TLV she should do:

If smpstate is not SMPSTATE_EXPECT2:
Set smpstate to SMPSTATE_EXPECT1 and send a type 6 TLV (SMP abort) to Bob.
If smpstate is SMPSTATE_EXPECT2:
Verify Bob's zero-knowledge proofs for g2b, g3b, Pb and Qb:
  1. Check that g2b, g3b, Pb and Qb are >= 2 and <= modulus-2.
  2. Check that c2 = SHA256(3, g1D2 g2bc2).
  3. Check that c3 = SHA256(4, g1D3 g3bc3).
  4. Check that cP = SHA256(5, g3D5 PbcP, g1D5 g2D6 QbcP).
Create a type 4 TLV (SMP message 3) and send it to Bob:
  1. Pick random exponents r4, r5, r6 and r7. These will be used to add a blinding factor to the final results, and to generate zero-knowledge proofs that this message was created honestly.
  2. Compute g2 = g2ba2 and g3 = g3ba3
  3. Compute Pa = g3r4 and Qa = g1r4 g2x
  4. Generate a zero-knowledge proof that Pa and Qa were created according to the protocol by setting cP = SHA256(6, g3r5, g1r5 g2r6), D5 = r5 - r4 cP mod q and D6 = r6 - x cP mod q.
  5. Compute Ra = (Qa / Qb) a3
  6. Generate a zero-knowledge proof that Ra was created according to the protocol by setting cR = SHA256(7, g1r7, (Qa / Qb)r7) and D7 = r7 - a3 cR mod q.
  7. Store the values of g3b, (Pa / Pb), (Qa / Qb) and a3 for use later in the protocol.
  8. Send Bob a type 4 TLV (SMP message 3) containing Pa, Qa, cP, D5, D6, Ra, cR and D7 in that order.
Set smpstate to SMPSTATE_EXPECT4.

Receiving a type 4 TLV (SMP message 3)

SMP message 3 is Alice's final message in the SMP exchange. It has the last of the information required by Bob to determine if x = y. It contains the following mpi values:

Pa, Qa
These values are used in the final comparison to determine if Alice and Bob share the same secret.
cP, D5, D6
A zero-knowledge proof that Pa and Qa were created according to the protcol given above.
Ra
This value is used in the final comparison to determine if Alice and Bob share the same secret.
cR, D7
A zero-knowledge proof that Ra was created according to the protcol given above.

When Bob receives this TLV he should do:

If smpstate is not SMPSTATE_EXPECT3:
Set smpstate to SMPSTATE_EXPECT1 and send a type 6 TLV (SMP abort) to Bob.
If smpstate is SMPSTATE_EXPECT3:
Verify Alice's zero-knowledge proofs for Pa, Qa and Ra:
  1. Check that Pa, Qa and Ra are >= 2 and <= modulus-2.
  2. Check that cP = SHA256(6, g3D5 PacP, g1D5 g2D6 QacP).
  3. Check that cR = SHA256(7, g1D7 g3acR, (Qa / Qb)D7 RacR).
Create a type 5 TLV (SMP message 4) and send it to Alice:
  1. Pick a random exponent r7. This will be used to generate Bob's final zero-knowledge proof that this message was created honestly.
  2. Compute Rb = (Qa / Qb) b3
  3. Generate a zero-knowledge proof that Rb was created according to the protocol by setting cR = SHA256(8, g1r7, (Qa / Qb)r7) and D7 = r7 - b3 cR mod q.
  4. Send Alice a type 5 TLV (SMP message 4) containing Rb, cR and D7 in that order.
Check whether the protocol was successful:
  1. Compute Rab = Rab3.
  2. Determine if x = y by checking the equivalent condition that (Pa / Pb) = Rab.
Set smpstate to SMPSTATE_EXPECT1, as no more messages are expected from Alice.

Receiving a type 5 TLV (SMP message 4)

SMP message 4 is Bob's final message in the SMP exchange. It has the last of the information required by Alice to determine if x = y. It contains the following mpi values:

Rb
This value is used in the final comparison to determine if Alice and Bob share the same secret.
cR, D7
A zero-knowledge proof that Rb was created according to the protcol given above.

When Alice receives this TLV she should do:

If smpstate is not SMPSTATE_EXPECT4:
Set smpstate to SMPSTATE_EXPECT1 and send a type 6 TLV (SMP abort) to Bob.
If smpstate is SMPSTATE_EXPECT4:
Verify Bob's zero-knowledge proof for Rb:
  1. Check that Rb is >= 2 and <= modulus-2.
  2. Check that cR = SHA256(8, g1D7 g3bcR, (Qa / Qb)D7 RbcR).
Check whether the protocol was successful:
  1. Compute Rab = Rba3.
  2. Determine if x = y by checking the equivalent condition that (Pa / Pb) = Rab.
Set smpstate to SMPSTATE_EXPECT1, as no more messages are expected from Bob.

User requests to begin SMP

If smpstate is not set to SMPSTATE_EXPECT1:
SMP is already underway. If you wish to restart SMP, send a type 6 TLV (SMP abort) to the other party and then proceed as if smpstate was SMPSTATE_EXPECT1. Otherwise, you may simply continue the current SMP instance.
If smpstate is set to SMPSTATE_EXPECT1:
No current exchange is underway. In this case, Alice should create a valid type 2 TLV (SMP message 1) as follows:
  1. Determine her secret input x, which is to be compared to Bob's secret y.
  2. Pick random values a2 and a3 (1536 bits). These will be Alice's exponents for the DH exchange to pick generators.
  3. Pick random values r2 and r3 (1536 bits). These will be used to generate zero-knowledge proofs that this message was created according to the protocol.
  4. Compute g2a = g1a2 and g3a = g1a3
  5. Generate a zero-knowledge proof that the exponent a2 is known by setting c2 = SHA256(1, g1r2) and D2 = r2 - a2 c2 mod q.
  6. Generate a zero-knowledge proof that the exponent a3 is known by setting c3 = SHA256(2, g1r3) and D3 = r3 - a3 c3 mod q.
  7. Store the values of x, a2 and a3 for use later in the protocol.
  8. Send Bob a type 2 TLV (SMP message 1) containing g2a, c2, D2, g3a, c3 and D3 in that order.
Set smpstate to SMPSTATE_EXPECT2.

User requests to abort SMP

In all cases, send a type 6 TLV (SMP abort) to the correspondent and set smpstate to SMPSTATE_EXPECT1.

Key Management

For each correspondent, keep track of:

Your two most recent DH public/private key pairs
our_dh[our_keyid] (most recent) and our_dh[our_keyid-1] (previous)
His two most recent DH public keys
their_y[their_keyid] (most recent) and their_y[their_keyid-1] (previous)

When starting a private conversation with a correspondent, generate two DH key pairs for yourself, and set our_keyid = 2. Note that all DH key pairs should have a private part that is at least 320 bits long.

When you send AKE messages:
Send the public part of our_dh[our_keyid-1], with the keyid field, of course, set to (our_keyid-1).
Upon completing the AKE:
If the specified keyid equals either their_keyid or their_keyid-1, and the DH pubkey contained in the AKE messages matches the one you've stored for that keyid, that's great. Otherwise, forget all values of their_y[], and of their_keyid, and set their_keyid to the keyid value given in the AKE messages, and their_y[their_keyid] to the DH pubkey value given in the AKE messages. their_y[their_keyid-1] should be set to NULL.
When you send a Data Message:
Set the sender keyid to (our_keyid-1), and the recipient keyid to (their_keyid). Set the DH pubkey in the Data message to the public part of our_dh[our_keyid]. Use our_dh[our_keyid-1] and their_y[their_keyid] to calculate session keys, as outlined below. Use the "sending AES key" to encrypt the message, and the "sending MAC key" to calculate its MAC.
When you receive a Data Message:
Use the keyids in the message to select which of your DH key pairs and which of his DH pubkeys to use to verify the MAC. If the keyids do not represent either the most recent key or the previous key (for either the sender or receiver), reject the message. Also reject the message if the sender keyid is their_keyid-1, but their_y[their_keyid-1] is NULL.

Otherwise, calculate the session keys as outlined below. Use the "receiving MAC key" to verify the MAC on the message. If it does not verify, reject the message.

Check that the counter in the Data message is strictly larger than the last counter you saw using this pair of keys. If not, reject the message.

If the MAC verifies, decrypt the message using the "receiving AES key".

Finally, check if keys need rotation:

Computing AES keys, MAC keys, and the secure session id

OTR uses Diffie-Hellman to calculate shared secrets in the usual way: if Bob knows x, and tells Alice gx, and Alice knows y, and tells Bob gy, then they each can calculate s = gxy: Alice calculates (gx)y, and Bob calculates (gy)x.

During the AKE, Alice and Bob each calculate s in this way, and then they each compute seven values based on s:

This is done in the following way:

c, m1, and m2 are used to create and verify the Reveal Signature Message; c', m1', and m2' are used to create and verify the Signature message.

If the user requests to see the secure session id, it should be displayed as two 32-bit bigendian unsigned values, in C "%08x" format. If the user transmitted the Reveal Signature message during the AKE that produced this ssid, then display the first 32 bits in bold, and the second 32 bits in non-bold. If the user transmitted the Signature message instead, display the first 32 bits in non-bold, and the second 32 bits in bold. This session id can be used by the parties to verify (say, over the telephone, assuming the parties recognize each others' voices) that there is no man-in-the-middle by having each side read his bold part to the other. [Note that this only needs to be done in the event that the users do not trust that their long-term signature keys have not been compromised.]

During the exchange of Data Messages, Alice and Bob use the keyids listed in the Data Message to select Diffie-Hellman keys to use to compute s, and the (4+len)-byte value of secbytes, as above.

From this, they calculate four values:

These keys are calculated as follows:

Extra symmetric key

OTR version 3 defines an additional symmetric key that can be derived by the communicating parties to use for application-specific purposes, such as file transfer, voice encryption, etc. When one party wishes to use the extra symmetric key, he or she creates a type 8 TLV attached to a Data Message (see above). The key itself is then derived using the same "secbytes" used to compute the encryption and MAC keys used to protect the Data Message. The extra symmetric key is derived by calculating h2(0xFF) and keeping the entire 256 bits, using the same definition of h2 as above.

Upon receipt of the Data Message containing the type 8 TLV, the recipient will compute the extra symmetric key in the same way. Note that the value of the extra symmetric key is not contained in the TLV itself.

Revealing MAC keys

Whenever you are about to forget either one of your old D-H key pairs, or one of your correspondent's old D-H public keys, take all of the receiving MAC keys that were generated by that key (note that there are up to two: the receiving MAC keys produced by the pairings of that key with each of two of the other side's keys; but note that you only need to take MAC keys that were actually used to verify a MAC on a message), and put them (as a set of concatenated 20-byte values) into the "Old MAC keys to be revealed" section of the next Data Message you send. This in done to allow the forgeability of OTR transcripts: once the MAC keys are revealed, anyone can modify an OTR message and still have it appear valid. But since we don't reveal the MAC keys until their corresponding pubkeys are being discarded, there is no danger of accepting a message as valid which uses a MAC key which has already been revealed.

Fragmentation

Some networks may have a maximum message size that is too small to contain an encoded OTR message. In that event, the sender may choose to split the message into a number of fragments. This section describes the format of the fragments. All OTR version 2 and 3 clients must be able to assemble received fragments, but performing fragmentation on outgoing messages is optional.

Transmitting Fragments
If you have information about the maximum size of message you are able to send (the different IM networks have different limits), you can fragment an encoded OTR message as follows:

Note that fragments are not themselves messages that can be fragmented: you can't fragment a fragment.

Receiving Fragments:
If you receive a message containing "?OTR|" (note that you'll need to check for this _before_ checking for any of the other "?OTR:" markers):

After this, if N > 0 and K == N, treat F as the received message.

If you receive a non-OTR message, or an unfragmented message, forget any stored fragment you may have, store "" as F and store (0,0) as (K,N).

OTR version 2 fragmented messages follow the same behaviour as described above, but do not list the sender and receiver instance tags.

For example, here is a Data Message we would like to transmit over a network with an unreasonably small maximum message size:

?OTR:AAMDJ+MVmSfjFZcAAAAAAQAAAAIAAADA1g5IjD1ZGLDVQEyCgCyn9hb
rL3KAbGDdzE2ZkMyTKl7XfkSxh8YJnudstiB74i4BzT0W2haClg6dMary/jo
9sMudwmUdlnKpIGEKXWdvJKT+hQ26h9nzMgEditLB8vjPEWAJ6gBXvZrY6ZQ
rx3gb4v0UaSMOMiR5sB7Eaulb2Yc6RmRnnlxgUUC2alosg4WIeFN951PLjSc
ajVba6dqlDi+q1H5tPvI5SWMN7PCBWIJ41+WvF+5IAZzQZYgNaVLbAAAAAAA
AAAEAAAAHwNiIi5Ms+4PsY/L2ipkTtquknfx6HodLvk3RAAAAAA==.

We could fragment this message into (for example) three pieces:

?OTR|5a73a599|27e31597,00001,00003,?OTR:AAMDJ+MVmSfjFZcAAAAA
AQAAAAIAAADA1g5IjD1ZGLDVQEyCgCyn9hbrL3KAbGDdzE2ZkMyTKl7XfkSx
h8YJnudstiB74i4BzT0W2haClg6dMary/jo9sMudwmUdlnKpIGEKXWdvJKT+
hQ26h9nzMgEditLB8v,
?OTR|5a73a599|27e31597,00002,00003,jPEWAJ6gBXvZrY6ZQrx3gb4v0
UaSMOMiR5sB7Eaulb2Yc6RmRnnlxgUUC2alosg4WIeFN951PLjScajVba6dq
lDi+q1H5tPvI5SWMN7PCBWIJ41+WvF+5IAZzQZYgNaVLbAAAAAAAAAAEAAAA
HwNiIi5Ms+4PsY/L2i,
?OTR|5a73a599|27e31597,00003,00003,pkTtquknfx6HodLvk3RAAAAAA
==.,

The protocol state machine

An OTR client maintains separate state for every correspondent. For example, Alice may have an active OTR conversation with Bob, while having an unprotected conversation with Charlie. This state consists of two main state variables, as well as some other information (such as encryption keys). The two main state variables are:

Message state

The message state variable, msgstate, controls what happens to outgoing messages typed by the user. It can take one of three values:

MSGSTATE_PLAINTEXT
This state indicates that outgoing messages are sent without encryption. This is the state that is used before an OTR conversation is initiated. This is the initial state, and the only way to subsequently enter this state is for the user to explicitly request to do so via some UI operation.
MSGSTATE_ENCRYPTED
This state indicates that outgoing messages are sent encrypted. This is the state that is used during an OTR conversation. The only way to enter this state is for the authentication state machine (below) to successfully complete.
MSGSTATE_FINISHED
This state indicates that outgoing messages are not delivered at all. This state is entered only when the other party indicates he has terminated his side of the OTR conversation. For example, if Alice and Bob are having an OTR conversation, and Bob instructs his OTR client to end its private session with Alice (for example, by logging out), Alice will be notified of this, and her client will switch to MSGSTATE_FINISHED mode. This prevents Alice from accidentally sending a message to Bob in plaintext. (Consider what happens if Alice was in the middle of typing a private message to Bob when he suddenly logs out, just as Alice hits Enter.)

Authentication state

The authentication state variable, authstate, can take one of four values (plus one extra for OTR version 1 compatibility):

AUTHSTATE_NONE
This state indicates that the authentication protocol is not currently in progress. This is the initial state.
AUTHSTATE_AWAITING_DHKEY
After Bob initiates the authentication protocol by sending Alice the D-H Commit Message, he enters this state to await Alice's reply.
AUTHSTATE_AWAITING_REVEALSIG
After Alice receives Bob's D-H Commit Message, and replies with her own D-H Key Message, she enters this state to await Bob's reply.
AUTHSTATE_AWAITING_SIG
After Bob receives Alice's D-H Key Message, and replies with his own Reveal Signature Message, he enters this state to await Alice's reply.
AUTHSTATE_V1_SETUP
For OTR version 1 compatibility, if Bob sends a version 1 Key Exchange Message to Alice, he enters this state to await Alice's reply.

After:

then, assuming the signature verifications succeed, the msgstate variable is transitioned to MSGSTATE_ENCRYPTED. Regardless of whether the signature verifications succeed, the authstate variable is transitioned to AUTHSTATE_NONE.

Policies

OTR clients can set different policies for different correspondents. For example, Alice could set up her client so that it speaks only OTR version 3, except with Charlie, who she knows has only an old client; so that it will opportunistically start an OTR conversation whenever it detects the correspondent supports it; or so that it refuses to send non-encrypted messages to Bob, ever.

The policies that can be set (on a global or per-correspondent basis) are any combination of the following boolean flags:

ALLOW_V1
Allow version 1 of the OTR protocol to be used (in general this document will not address OTR protocol version 1; see previous protocol documents for these details).
ALLOW_V2
Allow version 2 of the OTR protocol to be used.
ALLOW_V3
Allow version 3 of the OTR protocol to be used.
REQUIRE_ENCRYPTION
Refuse to send unencrypted messages.
SEND_WHITESPACE_TAG
Advertise your support of OTR using the whitespace tag.
WHITESPACE_START_AKE
Start the OTR AKE when you receive a whitespace tag.
ERROR_START_AKE
Start the OTR AKE when you receive an OTR Error Message.

Note that it is possible for UIs simply to offer the old "combinations" of options, and not ask about each one separately.

State transitions

There are twelve actions an OTR client must handle:

The following sections will outline what actions to take in each case. They all assume that at least one of ALLOW_V1, ALLOW_V2 or ALLOW_V3 is set; if not, then OTR is completely disabled, and no special handling of messages should be done at all. For version 1 messages, please refer to previous OTR protocol documents. For version 3 messages, someone receiving a message with a recipient instance tag specified that does not equal their own should discard the message and optionally warn the user. The exception here is the D-H Commit Message where the recipient instance tag may be 0, indicating that no particular instance is specified.

Receiving plaintext without the whitespace tag

If msgstate is MSGSTATE_PLAINTEXT:
Simply display the message to the user. If REQUIRE_ENCRYPTION is set, warn him that the message was received unencrypted.
If msgstate is MSGSTATE_ENCRYPTED or MSGSTATE_FINISHED:
Display the message to the user, but warn him that the message was received unencrypted.

Receiving plaintext with the whitespace tag

If msgstate is MSGSTATE_PLAINTEXT:
Remove the whitespace tag and display the message to the user. If REQUIRE_ENCRYPTION is set, warn him that the message was received unencrypted.
If msgstate is MSGSTATE_ENCRYPTED or MSGSTATE_FINISHED:
Remove the whitespace tag and display the message to the user, but warn him that the message was received unencrypted.

In any event, if WHITESPACE_START_AKE is set:

If the tag offers OTR version 3 and ALLOW_V3 is set:
Send a version 3 D-H Commit Message, and transition authstate to AUTHSTATE_AWAITING_DHKEY.
Otherwise, if the tag offers OTR version 2 and ALLOW_V2 is set:
Send a version 2 D-H Commit Message, and transition authstate to AUTHSTATE_AWAITING_DHKEY.

Receiving a Query Message

If the query message offers OTR version 3 and ALLOW_V3 is set:
Send a version 3 D-H Commit Message, and transition authstate to AUTHSTATE_AWAITING_DHKEY.
Otherwise, if the message offers OTR version 2 and ALLOW_V2 is set:
Send a version 2 D-H Commit Message, and transition authstate to AUTHSTATE_AWAITING_DHKEY.

Receiving an Error Message

Display the message to the user. If ERROR_START_AKE is set, reply with a Query Message.

User requests to start an OTR conversation

Send an OTR Query Message to the correspondent.

Receiving a D-H Commit Message

If the message is version 2 and ALLOW_V2 is not set, ignore this message. Similarly if the message is version 3 and ALLOW_V3 is not set, ignore the message. Otherwise:

If authstate is AUTHSTATE_NONE:
Reply with a D-H Key Message, and transition authstate to AUTHSTATE_AWAITING_REVEALSIG.
If authstate is AUTHSTATE_AWAITING_DHKEY:
This is the trickiest transition in the whole protocol. It indicates that you have already sent a D-H Commit message to your correspondent, but that he either didn't receive it, or just didn't receive it yet, and has sent you one as well. The symmetry will be broken by comparing the hashed gx you sent in your D-H Commit Message with the one you received, considered as 32-byte unsigned big-endian values.
If yours is the higher hash value:
Ignore the incoming D-H Commit message, but resend your D-H Commit message.
Otherwise:
Forget your old gx value that you sent (encrypted) earlier, and pretend you're in AUTHSTATE_NONE; i.e. reply with a D-H Key Message, and transition authstate to AUTHSTATE_AWAITING_REVEALSIG.
If authstate is AUTHSTATE_AWAITING_REVEALSIG:
Retransmit your D-H Key Message (the same one as you sent when you entered AUTHSTATE_AWAITING_REVEALSIG). Forget the old D-H Commit message, and use this new one instead. There are a number of reasons this might happen, including:
If authstate is AUTHSTATE_AWAITING_SIG or AUTHSTATE_V1_SETUP:
Reply with a new D-H Key message, and transition authstate to AUTHSTATE_AWAITING_REVEALSIG.

Receiving a D-H Key Message

If the message is version 2 and ALLOW_V2 is not set, ignore this message. Similarly if the message is version 3 and ALLOW_V3 is not set, ignore this message. Otherwise:

If authstate is AUTHSTATE_AWAITING_DHKEY:
Reply with a Reveal Signature Message and transition authstate to AUTHSTATE_AWAITING_SIG.
If authstate is AUTHSTATE_AWAITING_SIG:
If this D-H Key message is the same the one you received earlier (when you entered AUTHSTATE_AWAITING_SIG):
Retransmit your Reveal Signature Message.
Otherwise:
Ignore the message.
If authstate is AUTHSTATE_NONE, AUTHSTATE_AWAITING_REVEALSIG, or AUTHSTATE_V1_SETUP:
Ignore the message.

Receiving a Reveal Signature Message

If the message is version 2 and ALLOW_V2 is not set, ignore this message. Similarly if the message is version 3 and ALLOW_V3 is not set, ignore the message. Otherwise:

If authstate is AUTHSTATE_AWAITING_REVEALSIG:
Use the received value of r to decrypt the value of gx received in the D-H Commit Message, and verify the hash therein. Decrypt the encrypted signature, and verify the signature and the MACs. If everything checks out: Otherwise, ignore the message.
If authstate is AUTHSTATE_NONE, AUTHSTATE_AWAITING_DHKEY, AUTHSTATE_AWAITING_SIG, or AUTHSTATE_V1_SETUP:
Ignore the message.

Receiving a Signature Message

If the message is version 2 and ALLOW_V2 is not set, ignore this message. Similarly if the message is version 3 and ALLOW_V3 is not set, ignore the message. Otherwise:

If authstate is AUTHSTATE_AWAITING_SIG:
Decrypt the encrypted signature, and verify the signature and the MACs. If everything checks out: Otherwise, ignore the message.
If authstate is AUTHSTATE_NONE, AUTHSTATE_AWAITING_DHKEY, or AUTHSTATE_AWAITING_REVEALSIG:
Ignore the message.

User types a message to be sent

If msgstate is MSGSTATE_PLAINTEXT:
If REQUIRE_ENCRYPTION is set:
Store the plaintext message for possible retransmission, and send a Query Message.
Otherwise:
If SEND_WHITESPACE_TAG is set, and you have not received a plaintext message from this correspondent since last entering MSGSTATE_PLAINTEXT, attach the whitespace tag to the message. Send the (possibly modified) message as plaintext.
If msgstate is MSGSTATE_ENCRYPTED:
Encrypt the message, and send it as a Data Message. Store the plaintext message for possible retransmission.
If msgstate is MSGSTATE_FINISHED:
Inform the user that the message cannot be sent at this time. Store the plaintext message for possible retransmission.

Receiving a Data Message

If msgstate is MSGSTATE_ENCRYPTED:
Verify the information (MAC, keyids, ctr value, etc.) in the message.
If the verification succeeds:
  • Decrypt the message and display the human-readable part (if non-empty) to the user.
  • Update the D-H encryption keys, if necessary.
  • If you have not sent a message to this correspondent in some (configurable) time, send a "heartbeat" message, consisting of a Data Message encoding an empty plaintext. The heartbeat message should have the IGNORE_UNREADABLE flag set.
  • If the received message contains a TLV type 1, forget all encryption keys for this correspondent, and transition msgstate to MSGSTATE_FINISHED.
Otherwise, inform the user that an unreadable encrypted message was received, and reply with an Error Message.
If msgstate is MSGSTATE_PLAINTEXT or MSGSTATE_FINISHED:
Inform the user that an unreadable encrypted message was received, and reply with an Error Message.

User requests to end an OTR conversation

If msgstate is MSGSTATE_PLAINTEXT:
Do nothing.
If msgstate is MSGSTATE_ENCRYPTED:
Send a Data Message, encoding a message with an empty human-readable part, and TLV type 1. Transition msgstate to MSGSTATE_PLAINTEXT.
If msgstate is MSGSTATE_FINISHED:
Transition msgstate to MSGSTATE_PLAINTEXT.