Skip to content
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
289 lines (221 sloc) 11 KB

PubSub interface for libp2p

Generalized publish/subscribe interface for libp2p.

Lifecycle Stage Maturity Status Latest Revision
3A Recommendation Active r2, 2019-02-01

Authors: @whyrusleeping

Interest Group: @yusefnapora, @raulk, @vyzo, @Stebalien, @jamesray1, @vasco-santos

See the lifecycle document for context about maturity level and spec status.

Table of Contents


This is the specification for generalized pubsub over libp2p. Pubsub in libp2p is currently still experimental and this specification is subject to change. This document does not go over specific implementation of pubsub routing algorithms, it merely describes the common wire format that implementations will use.

libp2p pubsub currently uses reliable ordered streams between peers. It assumes that each peer is certain of the identity of each peer it is communicating with. It does not assume that messages between peers are encrypted, however encryption defaults to being enabled on libp2p streams.

You can find information about the PubSub research and notes in the following repos:



All communication between peers happens in the form of exchanging protobuf RPC messages between participating peers.

The RPC protobuf is as follows:

message RPC {
	repeated SubOpts subscriptions = 1;
	repeated Message publish = 2;

	message SubOpts {
		optional bool subscribe = 1;
		optional string topicid = 2;

This is a relatively simple message containing zero or more subscription messages, and zero or more content messages. The subscription messages contain a topicid string that specifies the topic, and a boolean signifying whether to subscribe or unsubscribe to the given topic. True signifies 'subscribe' and false signifies 'unsubscribe'.

The Message

The RPC message can contain zero or more messages of type 'Message' (perhaps this should be named better?). The Message protobuf looks like this:

message Message {
	optional string from = 1;
	optional bytes data = 2;
	optional bytes seqno = 3;
	repeated string topicIDs = 4;
	optional bytes signature = 5;
	optional bytes key = 6;

The from field denotes the author of the message, note that this is not necessarily the peer who sent the RPC this message is contained in. This is done to allow content to be routed through a swarm of pubsubbing peers.

The data field is an opaque blob of data, it can contain any data that the publisher wants it to.

The seqno field is a 64-bit big-endian uint that is a linearly increasing number that is unique among messages originating from each given peer. No two messages on a pubsub topic from the same peer should have the same seqno value, however messages from different peers may have the same sequence number, so this number alone cannot be used to address messages. Notably the 'timecache' in use by the go implementation contains a message_id, which is constructed from the concatenation of the seqno and the from fields. This message_id is then unique among messages. It was also proposed in #116 to use a message_hash, however, it was noted: "a potential caveat with using hashes instead of seqnos: the peer won't be able to send identical messages (e.g. keepalives) within the timecache interval, as they will get rejected as duplicates."

The topicIDs field specifies a set of topics that this message is being published to.

The signature and key fields are used for message signing, as explained below.

The size of the Message should be limited, say to 1 MiB, but could also be configurable, for more information see issue 118, while messages should be rejected if they are over this size. Note that for applications where state such as messages is stored, such as blockchains, it is suggested to have some kind of storage economics (see e.g. here, here and here).

Message Signing

Messages can be optionally signed, and it is up to the peer whether to accept and forward unsigned messages.

For signing purposes, the signature and key fields are used:

  • The signature field contains the signature.
  • The key field contains the signing key when it cannot be inlined in the source peer ID. When present, it must match the peer ID.

The signature is computed over the marshalled message protobuf excluding the key field. The protobuf blob is prefixed by the string libp2p-pubsub: before signing.

When signature validation fails for a signed message, the implementation must drop the message and omit propagation. Locally, it may treat this event in whichever manner it wishes (e.g. logging).

The Topic Descriptor

The topic descriptor message is used to define various options and parameters of a topic. It currently specifies the topic's human readable name, its authentication options, and its encryption options. The AuthOpts and EncOpts of the topic descriptor message are not used in current implementations, but may be used in future. For clarity, this is added as a comment in the file, and may be removed once used.

The TopicDescriptor protobuf is as follows:

message TopicDescriptor {
	optional string name = 1;
	// AuthOpts and EncOpts are unused as of Oct 2018, but
	// are planned to be used in future.
	optional AuthOpts auth = 2;
	optional EncOpts enc = 3;

	message AuthOpts {
		optional AuthMode mode = 1;
		repeated bytes keys = 2;

		enum AuthMode {
			NONE = 0;
			KEY = 1;
			WOT = 2;

	message EncOpts {
		optional EncMode mode = 1;
		repeated bytes keyHashes = 2;

		enum EncMode {
			NONE = 0;
			WOT = 2;

The name field is a string used to identify or mark the topic. It can be descriptive or random or anything that the creator chooses.

Note that instead of using, for privacy reasons the TopicDescriptor struct may be hashed, and used as the topic ID. Another option is to use a CID as a topic ID. While a consensus has not been reached, for forwards and backwards compatibility, using an enum TopicID that allows custom types in variants (i.e. Name, hashedTopicDescriptor, CID) may be the most suitable option if it is available within an implementation's language (otherwise it would be implementation defined).

The auth field specifies how authentication will work for this topic. Only authenticated peers may publish to a given topic. See 'AuthOpts' below for details.

The enc field specifies how messages published to this topic will be encrypted. See 'EncOpts' below for details.


The AuthOpts message describes an authentication scheme. The mode field specifies which scheme to use, and the keys field is an array of keys. The meaning of the keys field is defined by the selected AuthMode.

There are currently three options defined for the AuthMode enum:

AuthMode 'NONE'

No authentication, anyone may publish to this topic.

AuthMode 'KEY'

Only peers whose peerIDs are listed in the keys array may publish to this topic, messages from any other peer should be dropped.

AuthMode 'WOT'

Web Of Trust: any trusted peer may publish to the topic. A trusted peer is one whose peerID is listed in the keys array, or any peer who is 'trusted' by another trusted peer. The mechanism of signifying trust in another peer is yet to be defined.


The EncOpts message describes an encryption scheme for messages in a given topic. The mode field denotes which encryption scheme will be used, and the keyHashes field specifies a set of hashes of keys whose purpose may be defined by the selected mode.

There are currently three options defined for the EncMode enum:

EncMode 'NONE'

Messages are not encrypted, anyone can read them.


Messages are encrypted with a preshared key. The salted hash of the key used is denoted in the keyHashes field of the EncOpts message. The mechanism for sharing the keys and salts is undefined.

EncMode 'WOT'

Web Of Trust publishing. Messages are encrypted with some certificate or certificate chain shared amongst trusted peers. (Spec writer's note: this is the least clearly defined option and my description here may be wildly incorrect, needs checking).

Topic Validation

Implementations MUST support attaching validators to topics.

Validators have access to the Message and can apply any logic to determine its validity. When propagating a message for a topic, implementations will invoke all validators attached to that topic, and will only continue propagation if, and only if all, validations pass.

In its simplest form, a validator is a function with signature (peer.ID, *Message) => bool, where the return value is true if validation passes, and false otherwise.

Local handling of failed validation is left up to the implementation (e.g. logging).

Implementations MAY allow dynamically adding and removing validators at runtime.

You can’t perform that action at this time.