Whispr

• Name: Whispr
• URL: https://github.com/whisprchat
• Category: decentralized messaging protocol / post-quantum sealed-sender chat stack / metadata-minimizing relay infrastructure
• Summary: Whispr is a privacy-heavy chat stack with a live architecture gap. The note matters because it keeps the relay, identity server, and authorship-hiding pieces separate instead of hiding everything under encrypted messenger copy. It also shows a familiar problem: the public README still describes an older pairwise-ratchet design, while the March 2026 whitepaper has moved on to a more ambitious APQ-MLS plus sealed-sender story.
• What it does:
  • Frames Whispr as a decentralized end-to-end encrypted chat system spread across separate protocol, app, relay, identity, and cryptography repositories under the whisprchat organization
  • Uses a desktop architecture where a React/Tauri client talks to a Rust sidecar that holds keys, manages protocol state, and handles relay and identity-server communication
  • Defines a relay role that stores and forwards sealed entries by topic and is meant to verify only membership proofs rather than learn message contents or authorship
  • Uses a separate identity server for registration, challenge-response authentication, profile data, composite public-key bundles, presence/status, device management, and MLS KeyPackage distribution
  • In the March 2026 whitepaper, replaces a simple one-layer chat design with APQ-MLS: a post-quantum MLS session for key refresh plus a traditional MLS session for application-message efficiency, linked by periodic PSK export
  • Adds sealed sender using BBS+ anonymous credentials, a hybrid ML-KEM-768 + X25519 transport handshake, SHA3-256 fingerprints over composite key bundles, and feature surfaces such as message editing, deletion, emoji reactions, typing indicators, direct messages, multi-relay federation, and a server-discovery directory
• Key claims:
  • The real analytical use here is not post-quantum chat branding. It is the explicit split between relay, identity server, and group-state distribution, which makes it easier to compare Whispr against cleaner transport notes like Waku or stronger metadata-privacy notes like Nym.
  • The whitepaper’s APQ-MLS split is the mechanism worth keeping: a heavier post-quantum MLS session refreshes key material, then feeds a lighter conventional MLS session that carries most day-to-day message traffic.
  • The relay story matters because Whispr is trying to hide authorship from the relay itself through sealed sender and BBS+ membership proofs, not just encrypt payloads in transit.
  • The identity server is still a real control point. Registration, authentication, profile management, composite key-bundle distribution, KeyPackage handling, and server discovery all sit there.
  • The maturity caveat is the note. Public source drift is obvious: the org README still describes the older PRGE-era stack, while the newer whitepaper specifies APQ-MLS, hybrid PQ transport, and anonymous credentials. Treat this as a mechanism note first, not proof of a mature live network.
• Whitepaper: Yes. The strongest primary artifact in this pass is the March 2026 protocol whitepaper source, supplemented by the organization README and the thin public docs surface; see ../whitepapers/whispr-primary-sources-2026-05-14.md.
• Sources:
  • https://github.com/whisprchat
  • https://github.com/whisprchat/.github/blob/main/README.md
  • https://github.com/whisprchat/whitepaper
  • https://raw.githubusercontent.com/whisprchat/whitepaper/main/whitepaper.tex
  • https://docs.whspr.one/

Internal linkages

• Keep this note reading upward toward waku, xmtp, and nym.

Control surface

• The leverage sits in identity-server policy, relay behavior, key-bundle distribution, and whichever protocol version actually ships.

• Keep the maturity caveat visible. Right now this is a useful mechanism note, not a transport anchor.

• Last reviewed: 2026-06-11 UTC