RetroShare

Open-source Friend-to-Friend networking, honestly reviewed. Not for everyone — and the project will tell you that itself.

TL;DR

• What it is: A decentralized, encrypted Friend-to-Friend (F2F) communication platform covering chat, mail, file sharing, forums, channels, and experimental VoIP — all running over a self-built mesh of trusted nodes [website][1].
• Who it’s for: Privacy-focused technical users who want communication infrastructure that no central authority can shut down, surveil, or monetize. Not for non-technical teams [1][3].
• Cost: Free software, no SaaS tier, no ads, no terms of service. Infrastructure cost is whatever server or desktop you run it on [website].
• Key strength: The deepest privacy model in the open-source communication category. F2F architecture means only your direct peers know your IP. Tor/I2P integration takes it further [website][2].
• Key weakness: The UI looks like 2008 and works like 2008. The setup requires exchanging certificates with each contact manually. A code audit found exploitable bugs. One widely-cited privacy community thread describes it as “too complicated” and “not mature enough” for real-world use [1][2][3].

---

What is RetroShare

RetroShare is a decentralized communication toolkit. The core idea is a Friend-to-Friend network: you generate a cryptographic identity, export a certificate, and exchange it manually with people you trust. Those people become your direct nodes. Services — chat, file sharing, forums, channels, mail — run over the encrypted mesh you and your network build together. Nobody hosts the infrastructure. No company holds your data [website][README].

The project was founded in 2006 by a developer named drbob as a platform for “secure communications and file sharing with friends.” It has 1,955 GitHub stars and sits at version 0.6.7a as of this review. The codebase is C++, cross-platform (Android, Linux, macOS, Windows), and genuinely active — commits as recent as April 2026 [1][website].

What separates it from Signal, Matrix, or XMPP is the topology. There are no servers to subpoena, no federation hubs to block, no phone numbers to tie to identities. The tradeoff is that you are the infrastructure. If your friends aren’t online, asynchronous delivery works through a store-and-forward mechanism using friends’ nodes. If you want to talk to someone new, you need a mutual introduction through the network, or to meet them somewhere and exchange certificates [website][2].

The project’s own stated goals are explicit about the threat model: evade intelligence agencies, resist censorship, maintain freedom of speech. This is not a Slack replacement for a marketing team. It is a communication layer built for people who have thought seriously about who their adversaries are [website].

---

Why people choose it

The reviews we found don’t come from happy non-technical users — they come from the privacy and security community, and they’re mixed.

The case for RetroShare is stated cleanly by the Awesome Privacy project [1]: “Secure group communications, with the option to be used over Tor or I2P.” The EFF’s secure messaging scorecard gave it 6 out of 7 points in 2014 — losing only for lack of an independent code audit (that audit has since happened, with mixed results) [2]. The features are genuinely comprehensive: encrypted 1:1 chat, decentralized forums, channel subscriptions, BitTorrent-style file transfer with anonymous tunnels, and even an experimental VoIP plugin [website].

The case against is louder in the community discussions. The Privacy Guides forum thread [3] — from a user who evaluated essentially every privacy-focused messenger — places RetroShare in a bucket with Tox and Berty: “too complicated.” The same thread dismisses the entire category of F2F networks as having use cases too narrow to gain real traction.

Whonix’s documentation [2] goes further, identifying three specific privacy problems even before you’ve opened the app:

1. You expose your social graph to a global passive adversary because friends connect to friends directly.
2. Your public IP is available in the DHT (distributed hash table), allowing location tracking.
3. Your visible username is exposed in the TLS certificate when someone connects to your node.

All three problems can be mitigated by disabling the DHT and routing RetroShare through Tor or I2P. RetroShare actually ships Tor-integrated builds specifically for this purpose. But this is a meaningful extra configuration step, not something that happens automatically [2][website].

The security audit by Elttam uncovered “many bugs in the code, some remotely exploitable” and concluded that the codebase “lacked secure coding practice.” Those bugs were reportedly fixed promptly. This is not a death sentence — open-source projects with active audits often come out stronger — but it is worth knowing before you build your threat-model around this software [2].

---

Features

Based on the README, website, and third-party descriptions:

Communication:

• 1:1 encrypted text and image chat [website]
• Decentralized multi-user chat rooms (comparable to IRC) [website]
• “Distant chat” — secure messaging with friends-of-friends via the mesh [website]
• Encrypted mail with store-and-forward for offline delivery [website]
• VoIP and video calling — labeled “Experimental Prototype” on the website itself [website]

Content and collaboration:

• Decentralized forums with offline read/write (sync happens when connected) [website]
• Boards for sharing links and images with voting and comments [website]
• Channels for publishing files with subscription and auto-download [website]
• File sharing with BitTorrent-style swarming for large files [website]
• Anonymous tunnels for file transfers beyond direct friends [website]

Network and privacy:

• Friend-to-Friend (F2F) topology — only neighbors know your IP by default [website]
• Optional Tor and I2P integration, including pre-configured Tor builds [website][2]
• UDP, TCP, Tor, I2P transport support [website]
• UPnP/NAT-PMP for router traversal [website]
• DHT for friend discovery (optional — disabling improves privacy) [website][2]
• Authentication via PGP keys; transport encryption via TLS with Perfect Forward Secrecy [website]

Developer surface:

• JSON API for external integrations [website]
• Plugin architecture [website]
• GXS (Generic eXchange System) — the internal data distribution layer for forums, channels, and messaging [website]

Platform:

• Android, Linux, macOS, Windows [website]
• Actively maintained — commits from March and April 2026 visible on GitHub [1]

---

Pricing: SaaS vs self-hosted math

RetroShare has no commercial offering, no pricing page, and no SaaS tier. The software is free. You run it on your own hardware: a desktop, a laptop, a VPS, or a Raspberry Pi. There are no subscription fees, no per-user costs, no task limits [website].

The relevant comparison for someone thinking about self-hosted communication isn’t RetroShare vs. its own cloud tier — it’s RetroShare vs. the tools it could replace:

• Slack: $7.25/user/month (Pro). A 10-person team pays $87/month. A 50-person team: $435/month.
• Discord Nitro (server boosts): Various, but the platform is still centralized and ad-supported.
• Signal: Free, but centralized, requires phone numbers, and runs on Signal’s servers.
• Matrix/Element: Free to self-host, federated, no per-user cost but requires server maintenance.

If the question is purely “can I replace Slack billing with zero dollars,” RetroShare technically answers yes. But the honest answer is that the operational costs shift: you’ll spend time on certificate exchange with every new contact, managing Tor configuration if you want real privacy, and explaining to teammates why they need to compile software from a build guide [2][README].

For a solo privacy researcher or journalist communicating with a small trusted circle: the cost math works, and no other tool matches the privacy model. For a 10-person startup team that currently pays $87/month for Slack: RetroShare is not the tool. The friction cost alone, measured in lost productivity, will exceed the Slack bill within weeks.

---

Deployment reality check

This is where the honest answer diverges sharply from the marketing copy.

The README’s install path is not Docker Compose with a web UI. It is a compilation guide split by operating system: one file for Windows (MSYS2), one for macOS, one for Linux (Debian/Ubuntu). There are pre-built binaries via the website and an open build service for some distributions, but the Whonix documentation notes those builds “haven’t been updated in approximately two years” for some configurations, and “no packages are available for Debian Trixie / Whonix 18” [2].

What you actually need:

• A desktop or server running a supported OS
• Willingness to follow a platform-specific build guide or trust pre-built binaries
• At least one other person to exchange certificates with (there is no network without contacts)
• If you want real anonymity: a Tor setup, understanding of the DHT privacy issues, and patience

What can go sideways:

• Whonix flags that RetroShare packages are “signed with weak 1024-bit keys” and recommends using Ricochet IM with OnionShare instead as a safer alternative [2].
• The security audit found remotely exploitable bugs in the codebase. Fixes were applied, but the audit’s overall verdict on secure coding practices was negative [2].
• The DHT runs continuously by default and causes “continuous CPU usage” — relevant if you’re running this on a low-power device [2].
• VoIP is explicitly labeled “Experimental Prototype” on the project’s own homepage. Do not deploy this expecting working calls [website].
• The bootstrap problem is real: the network is only as useful as the number of people you can get to join it. Convincing five colleagues to exchange PGP-derived certificates is a different ask than “download Signal and send me your number.”

Realistic time estimate for a technical user: 2–4 hours for initial setup and first successful connection. For a non-technical user without a guide tailored to their OS: getting stuck is the expected outcome, not the edge case [1][README].

---

Pros and cons

Pros

• No central point of failure or control. No company can shut it down, change terms of service, sell data, or respond to a subpoena with user communications [website][1].
• Deep privacy model. F2F means only direct contacts know your IP. Tor integration removes even that. No phone number, no email required [website][2].
• Genuinely comprehensive feature set. Chat, mail, file sharing, forums, channels, boards, VoIP — in a single application. Most privacy tools offer one or two of these [website][1].
• Offline-tolerant. Mail and forum posts work offline and sync when you reconnect [website].
• Tor/I2P native. RetroShare ships dedicated builds with Tor embedded and auto-managed [website][2].
• Active development. Commits in April 2026. Not abandoned [1].
• No ads, no analytics, no terms of service. The project makes this explicit [website].

Cons

• The UI is genuinely dated. Awesome Privacy describes it as “quite retro” and notes it “may not be appropriate for a non-technical team” [1]. This is an understatement.
• Complex to use. The Privacy Guides community thread places it in the “too complicated” category outright, alongside tools it dismisses as not ready for real users [3].
• Bootstrap problem. You need contacts who are also running RetroShare to have anyone to communicate with. Cold-starting a network is hard [website].
• Weak package signing. Whonix flags 1024-bit signing keys as a security concern and recommends against installing via third-party repos [2].
• Security audit found exploitable bugs. The Elttam audit concluded the codebase lacks secure coding practice. Bugs were fixed, but the systemic finding stands [2].
• DHT leaks IP by default. The default configuration exposes your IP in the DHT and username in TLS certificates. Requires active configuration to mitigate [2].
• Package availability gaps. No packages for Debian Trixie/Whonix 18 at time of writing [2].
• VoIP is experimental. The website uses the words “Experimental Prototype.” Take that literally [website].
• Not independently audited for cryptographic correctness. The EFF scorecard noted this in 2014; the subsequent audit focused on code bugs, not cryptographic design [2].

---

Who should use this / who shouldn’t

Use RetroShare if:

• You’re a journalist, activist, or researcher who needs communication infrastructure that resists government-level surveillance and can’t rely on any central service.
• You’re building a private network with a small group of technically capable, highly trusted contacts.
• You want a single application that covers chat, file sharing, forums, and mail in a decentralized setup.
• You’re comfortable compiling software or following a multi-step install guide and troubleshooting what breaks.
• You’ve read the Whonix page, understood the privacy caveats, and configured Tor accordingly [2].

Skip it (consider Signal or SimpleX instead) if:

• You want private messaging with a good UX that non-technical contacts will actually use. Signal covers most people’s threat model and requires nothing more than a phone number. SimpleX removes even the phone number requirement and received positive mentions in the same community thread that dismisses RetroShare [3].
• You’re a founder looking to replace Slack or Discord for team communication. The bootstrap friction alone rules this out for most teams.

Skip it (consider Matrix/Element instead) if:

• You want decentralized, self-hosted communication with a modern UI and a growing ecosystem. Matrix has federation, a real client (Element), and doesn’t require manual certificate exchange for every contact.

Skip it (consider Briar instead) if:

• Your threat model requires working over Bluetooth and WiFi without any internet. Briar is purpose-built for this; RetroShare is not.

---

Alternatives worth considering

• SimpleX Chat — no user identifiers at all (no phone, no username as identifier), modern UI, actively developed. The Privacy Guides community thread cited above calls it one of only two messengers “worth considering” [3].
• Session — no phone number required, onion-routed, decent mobile apps. Less cryptographically ambitious than SimpleX but easier to onboard contacts [3].
• Matrix/Element — federated, self-hostable, modern UI, E2E encryption in rooms. Requires a server but the setup is well-documented and the ecosystem is large.
• Briar — fully peer-to-peer, works over Tor, Bluetooth, and WiFi. Designed specifically for activists and journalists in high-surveillance environments. More limited feature set than RetroShare but simpler and more audited.
• Tox — another F2F-style decentralized messenger; the Privacy Guides thread describes it as “not well maintained” [3], putting it in a similar boat to RetroShare.
• Jami — P2P, no central server, cross-platform, more modern UI than RetroShare. The same thread calls it “inconsistent in my experience” [3].

For a non-technical founder who wants to move off a corporate communication platform: none of these are the right answer, including RetroShare. Matrix with managed hosting (like Element One) is the closest to “decentralized communication without the pain.”

---

Bottom line

RetroShare is a serious, long-running open-source project that solves a specific and difficult problem: building a private, censorship-resistant communication layer with no central infrastructure. For that problem, it genuinely works, and nothing else in the space covers the same combination of features — chat, mail, file sharing, forums, and channels — in a single self-contained application with Tor integration.

The honest caveat is everything else. The UI looks like 2008. A security audit found exploitable bugs. Package signing uses weak keys. The default configuration leaks your IP via DHT. VoIP is labeled “Experimental Prototype” on the project’s own website. The community most likely to appreciate its privacy model explicitly describes it as “too complicated.” And the bootstrap problem — you need contacts who also run RetroShare — means the network is only as useful as the circle of people you can convince to join.

If you’re a non-technical founder looking to cut a SaaS communication bill, this is not the answer. If you’re a journalist, researcher, or activist building a private mesh with a trusted circle of technical contacts who have thought carefully about their threat model — and who have read the Whonix configuration guide — RetroShare is one of the more architecturally honest tools in this space.

---

Sources

1. Awesome Privacy — RetroShare (Communication / Team Collaboration). https://awesome-privacy.xyz/communication/team-collaboration/retroshare/
2. Whonix Wiki — RetroShare (Installation, privacy caveats, and security audit notes). https://www.whonix.org/wiki/RetroShare
3. Privacy Guides Community Forum — “Which private messaging / communication app is best?” (Dec 2024). https://discuss.privacyguides.net/t/which-private-messaging-communication-app-is-best/23335

Primary sources:

• GitHub repository and README: https://github.com/retroshare/retroshare (1,955 stars, C++, active as of April 2026)
• Official website: https://retroshare.cc
• Downloads and current version (0.6.7a): https://retroshare.cc/downloads.html