I discovered Radicle back in 2020 (when their website looked incredible: https://web.archive.org/web/20201201030505/https://radicle.x...). I bounced off of it, in part due to being unable to effectively delete repositories. They used to have an FAQ about that—looks like it's gone now, though the public-private repository area is much more fleshed out (you can make a repo private, in which case no new updates will be publicized but the history will still exist). In truth, it's just profoundly difficult to effectively "delete" things in a decentralized system (see: Matrix, BitTorrent, et. al.). But definitely something to consider; people accidentally upload secrets, and want to have some recourse when that happens.
Still, time has passed and I have become more interested in GitHub alternatives (https://figbert.com/posts/ideating-tragit/). Will likely end up moving to Tangled. But first I need to add support over there for pushing over HTTPS...
In general, whatever has been made public, is hard to unmake public. There can always remain a copy.
It is acutely true for peer-to-peer distributed systems like Radicle, but is still true for the more centralized GitHub, and for the Web in general. If you want to be able to walk something back, better don't publush it.
Publishing a token or an ssh key should not be a big deal, such a token or key should be immediately revoked, which is as good as deleted. The problem occurs when the fact is not noticed immediately. A git hook can help avoid such mishaps.
It’s a fair point about the difficulty in deleting data in a decentralised system but that’s also true in more centralised systems like GitHub or any other website. Once some data is out there, you have no control over whether it can be removed. Other parties could have copied it and may re-share it. “All” removing data from a centralised system does is slow down this spread, sometimes to the point that it is effectively deleted but that can’t be guaranteed.
The best response I’m aware of is to invalidate any secrets that have been accidentally shared. Sometimes that is easier said than done though. And of course there are plenty of other reasons someone might want to delete data.
Love to see it. Unlike tangled.org this is local-first and has a solid story around private repos. I’m bullish on distributed forges in general, but I’m all for experimentation in figuring out exactly what that looks like.
AD: Feel free to post on our Zulip [^1] about your experiences of agentic workflow if you haven't already! Some of the team are interested in developing the agentic workflow experience.
Zulip used to have a really neat feature where you could open up certain channels to be indexable by search engines. Basically turning it into an open forum. The feature is still sorta around but completely broken. Quite unfortunate imo. It was a really neat solution to the walled garden problem with team chats
Zulip product lead here -- I'm not aware of anything in this space that used to work and no longer does.
There's long been a way to make Zulip channels searchable with zulip-archive (https://github.com/zulip/zulip-archive), and that continues to be available.
I wish they would make local-only deployment easier. For example, lets take 3 machines and try to setup Radicle to work only on those, without joining the common Radicle network. Like on-premises GitLab, but decentralized, without the need of the server. It requires quite some serious scripting and usecase not covered in the documentation.
Same here... I have been looking at all the self hosting options for a while but swapping one centralized system for another isn't what I would prefer. Running/using local and community gitea/forgejo has been good, but I hope decentralized radicle could be the solution to more independence.
This is a good fit primarily if you want to run a Radicle node that only seeds repos you tell it to. If you want to write code which you publish on Radicle, you need the tool that signs all your work with your private key (of your Radicle identity) - i.e. the `rad` CLI - and running that in a container isn't very useful. (e.g. think about how you'd replace `rad clone`). Having said that, here's a container image I maintain, in case it helps: https://quay.io/repository/radicle_garden/radicle-node
The story for sensitive/proprietary repositories doesn't feel well-fleshed-out yet:
> Radicle supports private repositories that are only shared among a trusted set of peers, not the entire network. These are not encrypted at rest but rely on selective replication and are thus completely invisible to the rest of the network.
There's no structural separation of public / private repositories; this is one bug or fat-finger away from a leak.
I was looking into how to get a radicle node running, but connected to only my own devices.
You can't seem to start the radicle node without it automatically connecting to the (generously provided) global shared seed nodes, and you can't examine/change the config without starting the node (at least with rad subcommands).
If you do `rad auth` and then delete the four seed addresses from ~/.radicle/config.json before starting the rad node, it helpfully re-adds them on startup:
[...]
INFO node Opening node database..
INFO node Address book is empty. Adding bootstrap nodes..
INFO node 4 nodes added to address book
[...]
(It doesn't add the bootstrap nodes if I instead add a seed "<randomDID>@10.254.254.254:8776".)
Is there a guide to using Radicle like one might use Fossil within a small company / within a small group of people (disconnected from the iris/rosa radicle.network seeds)?
When a repository is initialised as private [1], it's encoded in the repository identity document. The only way to change a repository to be public is to update the identity document; in the case of a repository with multiple delegates (repo maintainers in Radicle nomenclature), such a change requires a quorum to be met. So I'd say it's more than a fat-finger away from leaking.
> I was looking into how to get a radicle node running, but connected to only my own devices.
There's an open proposal [2] to introduce the concept of network configuration that would help with these kinds of use cases. Moreover, we're working on other ways to improve the collaboration experience for small teams, so stay tuned and thank you for this feedback!
In a nutshell, I summarise it like this: I see Open Source Software is a public good. As a public good, it doesn't belong in any proprietary platform, nor should orgs of any kind be in a position to gate keep it. We should rather host it on a public, peer-to-peer network that everyone can have access to.
Centralization is not a small problem to solve, and this looks like it might have done just that. More than censorship, I think the worst problem with github is long-term viability / reliability: they could pull the plug tomorrow and most people would likely lose everything but the raw code.
1. yes, it's in the user guide [1]
2. Git LFS should work, yes. Care to try and report back ? ;) Open a new topic on the Zulip #Support channel [2] if you run into any issues.
Regarding 2, we're actively working on radicle-artifact as an alternative to Git LFS with transport agnostic distribution (https/p2p), attestations, and redactions.
I saw https://radicle.dev/2025/08/14/jujutsu-with-radicle at the bottom of the homepage and assumed it meant Radicle was planning to support a native jj protocol to decentralize jj repositories. That's unfortunately not the case.
Is there a plan for jj repositories in radicle proper? I'm very tired of git shortcomings.
How do these federated forges deal with spam? If merge requests and issues are federated, does that mean that anybody running a radicle node (or interacting with one) can open issues or merge requests on all the repositories that you've made public? Or is there a whitelist (or something fancier?) to allow interaction only with specified nodes?
yes, we currently have just the base building blocks (`rad follow` - allowlist / `rad block` - blocklist) for the fancier things to be built on. When you "seed" a repository, by default you seed it only with "followed" scope, which means you would only see issues and "Patches" (our term for PR/MR) from the repo maintainers + other peers you follow (i.e. in your allowlist).
The more I have been using git and building my own tooling and services around it for usage, I have figured out that something like radicle feels like the right/better solution, definitely better than what github is atm.
There are rough edges and the seeding thing is a bit mehhh. And honestly there are a bunch of things I would do differently but I like the spirit of things.
Not sure where the authors of the project stand, but it's fun to see them make progress.
Prioritize discoverability, and self hosting on my own domain, with more customization for the protocol and ui/ux aspects.
Basically allow people to deploy something closer to gitea on a webserver initially, which has all the basic protocol features and sharing aspects implemented.
Currently the focus seems to have been very much on just the seeding & p2p part which honestly is good but not very useful for me as a developer/user who wants to work on a collaborative project.
Prioritizing P2P makes sense, but the UX/DX of radicle is not intuitive to put it mildly.
I went to the explorer radicle.network and my own as well and I still had no idea how to create an issue or submit a patch or login to an account or anything of that sort.
I opened the cli and figured out how to do all that, but then I created another node on my web server and now I have no idea how to connect to it, then I figured out how to do that, now how do I share my node without sharing it publicly, figured that out.
Cool now we are at the starting point, so every time I need to work on a project i will have to clone it, but why can't I create stuff in webui.
Maybe there is a setting I can't seem to find it yet.
But the jumping through hoops is mindbending. For a solution that could replace github for me and my friends who maintain a few public projects that wants small contributions from large number of folks as we try and maintain the local "tech" clubs.
Getting low information folks to use within 10 mins, is the bar that I have set for any platform I can use to replace github.
Radicle isn't that yet! anyways.
But I like the direction, putting contributors, access list, issues and prs in the repo is a brilliant idea, maybe put agent plans in it too, agent sessions as well maybe...
I would go as far as to say, something like a radicle network is better than every other alternative, but the UX is just not there yet.
This is all good feedback. (I'm not a radicle dev btw, just happened to be evaluating it for my own project this week.)
Fwiw, I think this is a matter of UI/UX. I think radicle provides the foundation for everything you describe -- it is just the pr/patch & issue tracking substrate, and the p2p layer for federation.
To make an analogy, I think what you are saying is "Look, BitTorrent is great and all, but I want to be able to search for music and movies. I want IMDB with a download button." That's fair, but the problem BitTorrent is solving is more fundamental, and you can build Napster or ThePirateBay on top.
Radicle, like BitTorrent, is solving the transport layer.
One challenge with distributed systems is that deletion and rollback expectations differ from traditional centralized platforms. Even if a platform removes data, cached or replicated versions can still persist elsewhere.
I'd like to see radicle replace crates.io. I can't get over Rust's dependency on github/Microsoft, and I can't get over the lack of namespacing.
All you would need is cargo compatibility, and a trusted namespace that kept up with the metadata of the current contents of crates.io, right?
edit: I really, really like rust, and love basically all of their choices about the language, but I can't stand the feeling that I'm being tricked into an ecosystem dependent on one of the worst behaved companies in the world, and I can't stand that a lot of rust projects smell like GPL-washing.
That being said, git is GPL and radicle is MIT, so it feels like the same thing, but Github also ain't git. I prefer MIT to MS; if radicle gets important enough and decides to rubpull, there will inevitably be a Free fork anyway.
> crates.io is moving away from GitHub-only authentication
Crates.io has not moved away from Github-only authentication, and got into the habit of yelling at people who complained about it.
> crates.io's attachment to GitHub is a fact about crates.io specifically, not the Cargo crate registry protocol
Is this just trivia you wanted to share? I feel like I covered it in the second sentence of the comment you're replying to.
> Cargo's support for Git repositories is generic across Git and has nothing to do with GitHub specifically
I'm looking to compile Rust projects from the semi-standard commonly-used crates. I do not want a Github account.
> Radicle offers nothing to a crate registry that a Git remote doesn't
Radicle offers peer-to-peer hosting that does not require a Github account.
> none of this has anything to do with the GPL.
Radicle is a project being built in Rust that partially reimplements git. Git is GPL-licensed, Radicle is MIT-licensed.
> It feels like you're just listing off things you like and don't like aesthetically.
I am unconcerned about your feelings. What I was saying is that I would like a peer-to-peer hosted, namespaced code repository that mirrors (or replaces) crates.io, and I do not want a github account to be necessary to use it.
> They have nothing to do with each other structurally.
I have no idea what "they" is referring to in this sentence.
Does radicle have some way of storing binaries outside of the source tree? I know cargo compiles from source by default, but AFAIK it can (and does) download binaries as well.
AD: Theres a project `artifact-cob` [^1] that enables some of the functionality you are looking for. You can host artifacts off-network and make 'releases' that point to them.
This just describes a watcher service that kicks jobs off on an external CI system and logs the results? Not much more detail than that.
Gitlab and Github have pages and pages going over the domain language used to configure the job triggers. Jobs can trigger other jobs either in response to completion or as a dependency etc etc.
I would say these radicle-ci designs as they are now are actually quite rudimentary. That's perfectly fine for an early project but at this point I think you have to say that they won't have a CI system ready for quite some time.
it's a little of both: on the one hand we're working on CI integrations (through generic webhooks or CI-engine-specific adapters that basically implement the CI engine's API [1]), so you can keep on using your existing CI solution. On the other hand, yes, there is also work towards a new CI engine (Ambient
[2]), which aims to make it safe and secure to run CI on other people's code, which is important when working towards a distributed, community-ran, CI system. We should have more docs regarding the CI story up on radicle.dev in the next week or two.
> What is Radicle? How is it different from Git/GitHub?
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git. Unlike centralized platforms like GitHub, there is no single entity controlling the network or user data. Repositories are replicated across peers in a decentralized manner. Radicle is an alternative for people and organizations who want full control of their data and user experience, without compromising on the social aspects of collaboration platforms.
(Quote from their FAQ).
This isn't even trying to answer the titular question... None of them, actually.
So, what is Radicle? A platform built on Git? What does this mean? A platform for what? What is it for?
Why Git/GitHub are used as if they were the same category of things? There's not even an attempt at answering the "how is this different from Git?" question. What does it offer that Git doesn't? Wtf is "forge"?
Radicle is an alternative... to what? I believe I have full control of my data in my Git repository... why do I need an alternative with even more control? How will I have even more control?
* * *
Maybe whatever this software does is actually useful or even good, but the documentation can't be worse.
I agree it's not immediately clear how it works, although I think I understand the role it's intending to fill.
If you're not familiar with the distinction between git and github it could be even more confusing.
As soon as I hear decentralized I have lots of questions about the underlying protocols. Their protocol page helps a little but also uses terms I'm not familiar with like "gossip protocol".
It would be nice for there to be a page that motivates the project a bit more, ie. explaining the technical problems they are attempting to solve before enumerating the components of the complex system they've built.
Someone looked at git - a distributed version control system that already works, that has been working since 2005, that Linus Torvalds wrote in a fit of pique and spite, and that currently hosts approximately all of the world's source code - and said, "This is good, but what if we added BitTorrent?" And then, presumably after consuming substances that I cannot legally inquire about, they continued: "And what if we added the Bitcoin peer-to-peer protocol?" And then, reaching a crescendo of architectural ambition that would make Icarus say "maybe dial it back a little," they concluded: "And we should DEFINITELY add blockchain identity."
Notice how each additional sentence makes the previous sentence worse, like a turducken of solutions looking for problems, or a nesting doll where every layer is a different kind of sadness. This, I presume, is what happens when you have a hammer and a screwdriver and a chainsaw, and you decide that every problem would be better solved by using all three simultaneously while riding a unicycle.
But seeing as it already does exist, it's pretty awesome.
Then why do they bring Git into the picture? They are not comparing themselves to Git...
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git.
This is a word salad that means nothing... more than 99% of moderns software is built on Git in one way or another. Anything that is designed to be used by more than a single user could be arguably called a "collaboration platform". This description completely fails to describe anything useful about the program they are trying to describe. For instance, Git is a peer-to-peer code collaboration platform built on Git. And the same can be said about a huge number of programs that share very little in terms of purpose or application.
When someone writes a definition, s.a. you'd find in encyclopedia, the rule is that it has to (a) link to the broader category of things (b) specify in what way the subject is (mostly) unique in the category (a). If you give only (a), then the reader walks away wondering how is subject different from anything else in (a). If you only give (b), then the reader needs to guess (a), and if they fail, they may misattribute or simply abandon efforts to understand the subject.
This attempt at "definition" is the textbook example of forgetting the (b). It's something that a 10-12 y.o. could come up with... this is not what an adult should strive for.
> They are not. Github is a centralized collaboration platform built on git, and radicle is a peer-to-peer collaboration platform built on git.
Built on meaning the technology is using Git under the hood, not that it is developed using git.
Edit:
Breaking down the “word salad”:
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git.
Peer-to-peer: it functions with individual nodes on the network spreading state for tracking it without relying on a single entity or centralised service.
Code collaboration platform (forge): you use it not just to store code but provides a way to keep track of “patches” (their term for PRs) and issues, amongst other things, to enable multiple people to collaborate on a code base
Built on git: the technology runs on top of git insofar as not only is the VCS just git, but the issues, patches, etc are stored in git. So the project isn’t merely developed using git, but when running the tool yourself it’s still backing everything under git.
I did understand the question. I think you're somehow confusing "built on git" with "we used git while we built it." But that's honestly just a weird guess. I really can't figure out how you could be confused about any of this.
I hope the definitions shared in the sibling comment will help.
> Then why do they bring Git into the picture? They are not comparing themselves to Git...
I would find it very strange if Github didn't mention git. It would be equally strange if radicle didn't mention git. They both wrap git.
I discovered Radicle back in 2020 (when their website looked incredible: https://web.archive.org/web/20201201030505/https://radicle.x...). I bounced off of it, in part due to being unable to effectively delete repositories. They used to have an FAQ about that—looks like it's gone now, though the public-private repository area is much more fleshed out (you can make a repo private, in which case no new updates will be publicized but the history will still exist). In truth, it's just profoundly difficult to effectively "delete" things in a decentralized system (see: Matrix, BitTorrent, et. al.). But definitely something to consider; people accidentally upload secrets, and want to have some recourse when that happens.
Still, time has passed and I have become more interested in GitHub alternatives (https://figbert.com/posts/ideating-tragit/). Will likely end up moving to Tangled. But first I need to add support over there for pushing over HTTPS...
In general, whatever has been made public, is hard to unmake public. There can always remain a copy.
It is acutely true for peer-to-peer distributed systems like Radicle, but is still true for the more centralized GitHub, and for the Web in general. If you want to be able to walk something back, better don't publush it.
Publishing a token or an ssh key should not be a big deal, such a token or key should be immediately revoked, which is as good as deleted. The problem occurs when the fact is not noticed immediately. A git hook can help avoid such mishaps.
It’s a fair point about the difficulty in deleting data in a decentralised system but that’s also true in more centralised systems like GitHub or any other website. Once some data is out there, you have no control over whether it can be removed. Other parties could have copied it and may re-share it. “All” removing data from a centralised system does is slow down this spread, sometimes to the point that it is effectively deleted but that can’t be guaranteed.
The best response I’m aware of is to invalidate any secrets that have been accidentally shared. Sometimes that is easier said than done though. And of course there are plenty of other reasons someone might want to delete data.
What are you missing from Radicle today that would make you chose Tangled over it?
I would love to know the need behind pushing over HTTPS. Is the SSH protocol not good?
Love to see it. Unlike tangled.org this is local-first and has a solid story around private repos. I’m bullish on distributed forges in general, but I’m all for experimentation in figuring out exactly what that looks like.
Unfortunate that they aren't using the AGPL license. It will allow SAAS companies to Embrace, Extend, and Extinguish.
Indeed, this seems to be already planned. https://radicle.dev/faq "Radworks intends to offer services built on top of Radicle."
AD: Radicle is protocol driven, my hope as a contributor is it becomes infrastructure.
Radicle is really underrated, especially when working with agents. I find it a joy to use for my agentic workflows.
If there's purely an agentic forge one day, it's likely going to be a distributed one, with cryptographic identities and signed artifacts by default.
AD: Feel free to post on our Zulip [^1] about your experiences of agentic workflow if you haven't already! Some of the team are interested in developing the agentic workflow experience.
[^1]: https://radicle.zulipchat.com/
I am surprised you don't use RustChat[1]
[1] https://rustchat.io/
How does it compare to MatterMost [1]?
[1] https://mattermost.com/
It doesn't have these pesky message and user limits, unlike Mattermost[1].
[1] https://github.com/mattermost/mattermost/issues/34271
Gotcha! Though that is not for the self-hosted version IIUC?
But yea, rust-chat in a way seems like more of a proper open source project ...
For self-hosted version as well, unless you patch it.
Doesn't even come close to Zulip in terms of features.
Zulip used to have a really neat feature where you could open up certain channels to be indexable by search engines. Basically turning it into an open forum. The feature is still sorta around but completely broken. Quite unfortunate imo. It was a really neat solution to the walled garden problem with team chats
Zulip product lead here -- I'm not aware of anything in this space that used to work and no longer does.
There's long been a way to make Zulip channels searchable with zulip-archive (https://github.com/zulip/zulip-archive), and that continues to be available.
Making the app-native web-public channels accessible to search engines is a big technical project, tracked as https://github.com/zulip/zulip/issues/21881.
"AD" ?
His initials
Advertisement
What makes it a joy to use with agents?
Probably less issues with account management and rate limits.
I noticed that they moved to a different domain last month: https://radicle.dev/2026/04/23/domain-move
.xyz and similar domains are often blocked by ISPs due to higher than usual rates of spam from domains using that TLD.
Makes sense. Usually if you want a 99 cent registration, .xyz is whats available.
I wish they would make local-only deployment easier. For example, lets take 3 machines and try to setup Radicle to work only on those, without joining the common Radicle network. Like on-premises GitLab, but decentralized, without the need of the server. It requires quite some serious scripting and usecase not covered in the documentation.
AD: We've just been discussing this! There's an open RIP [^1] that we will be supplying feedback for soon.
[^1]: https://radicle.zulipchat.com/#narrow/channel/369876-RIPs/to...
Same here... I have been looking at all the self hosting options for a while but swapping one centralized system for another isn't what I would prefer. Running/using local and community gitea/forgejo has been good, but I hope decentralized radicle could be the solution to more independence.
A containerised deployment would (understatement warning!) be useful.
This is a good fit primarily if you want to run a Radicle node that only seeds repos you tell it to. If you want to write code which you publish on Radicle, you need the tool that signs all your work with your private key (of your Radicle identity) - i.e. the `rad` CLI - and running that in a container isn't very useful. (e.g. think about how you'd replace `rad clone`). Having said that, here's a container image I maintain, in case it helps: https://quay.io/repository/radicle_garden/radicle-node
Somehow never heard of this before. It's even packaged in NixOS so I'll probably set up an always-on node.
AD: Big shout to our community member Defelo for maintaining those! Come join in!
Somewhat related:
Show HN: Epiq – Distributed Git based issue tracker TUI
https://news.ycombinator.com/item?id=48155570
The story for sensitive/proprietary repositories doesn't feel well-fleshed-out yet:
> Radicle supports private repositories that are only shared among a trusted set of peers, not the entire network. These are not encrypted at rest but rely on selective replication and are thus completely invisible to the rest of the network.
There's no structural separation of public / private repositories; this is one bug or fat-finger away from a leak.
I was looking into how to get a radicle node running, but connected to only my own devices.
You can't seem to start the radicle node without it automatically connecting to the (generously provided) global shared seed nodes, and you can't examine/change the config without starting the node (at least with rad subcommands).
If you do `rad auth` and then delete the four seed addresses from ~/.radicle/config.json before starting the rad node, it helpfully re-adds them on startup:
(It doesn't add the bootstrap nodes if I instead add a seed "<randomDID>@10.254.254.254:8776".)Is there a guide to using Radicle like one might use Fossil within a small company / within a small group of people (disconnected from the iris/rosa radicle.network seeds)?
Radicle contributor here.
When a repository is initialised as private [1], it's encoded in the repository identity document. The only way to change a repository to be public is to update the identity document; in the case of a repository with multiple delegates (repo maintainers in Radicle nomenclature), such a change requires a quorum to be met. So I'd say it's more than a fat-finger away from leaking.
> I was looking into how to get a radicle node running, but connected to only my own devices.
There's an open proposal [2] to introduce the concept of network configuration that would help with these kinds of use cases. Moreover, we're working on other ways to improve the collaboration experience for small teams, so stay tuned and thank you for this feedback!
[1] https://radicle.dev/guides/user#initializing-a-private-repos... [2] https://radicle.network/nodes/iris.radicle.network/rad:z3trN...
Congrats!
Maybe I'm not the target audience for this, so pardon my ignorance when I ask what problem does this solve? Centralization and censorship?
In a nutshell, I summarise it like this: I see Open Source Software is a public good. As a public good, it doesn't belong in any proprietary platform, nor should orgs of any kind be in a position to gate keep it. We should rather host it on a public, peer-to-peer network that everyone can have access to.
Centralization is not a small problem to solve, and this looks like it might have done just that. More than censorship, I think the worst problem with github is long-term viability / reliability: they could pull the plug tomorrow and most people would likely lose everything but the raw code.
I have two questions:
1. Does Radicle also work over TOR? 2. Does Radicle support Git LFS and/or Git Annex?
1. yes, it's in the user guide [1] 2. Git LFS should work, yes. Care to try and report back ? ;) Open a new topic on the Zulip #Support channel [2] if you run into any issues.
[1] - https://radicle.dev/guides/user#4-embracing-the-onion [2] - https://radicle.zulipchat.com/#narrow/channel/369873-Support
Regarding 2, we're actively working on radicle-artifact as an alternative to Git LFS with transport agnostic distribution (https/p2p), attestations, and redactions.
https://radicle.network/nodes/iris.radicle.network/rad%3Az4V...
I saw https://radicle.dev/2025/08/14/jujutsu-with-radicle at the bottom of the homepage and assumed it meant Radicle was planning to support a native jj protocol to decentralize jj repositories. That's unfortunately not the case.
Is there a plan for jj repositories in radicle proper? I'm very tired of git shortcomings.
This is cool. I think it fits the Agentic model a lot better; wrangling PATs and repo permissions on GitHub is quite tedious.
How are folks wiring up CI/CD? Seems you want to trigger compute DAGs on the patches you receive, what’s the latest and greatest here?
The general idea is using the radicle-ci-broker [1] to hook into repo events, and trigger different adapters based on the specific CI/CD tool you use.
We'll have more docs on this very soon. Also see https://news.ycombinator.com/item?id=48153535 for more info
[1]: https://radicle.network/nodes/seed.radicle.garden/rad%3AzBNX...
How do these federated forges deal with spam? If merge requests and issues are federated, does that mean that anybody running a radicle node (or interacting with one) can open issues or merge requests on all the repositories that you've made public? Or is there a whitelist (or something fancier?) to allow interaction only with specified nodes?
yes, we currently have just the base building blocks (`rad follow` - allowlist / `rad block` - blocklist) for the fancier things to be built on. When you "seed" a repository, by default you seed it only with "followed" scope, which means you would only see issues and "Patches" (our term for PR/MR) from the repo maintainers + other peers you follow (i.e. in your allowlist).
radicle is awesome and Just Works from what I've tried of it
The more I have been using git and building my own tooling and services around it for usage, I have figured out that something like radicle feels like the right/better solution, definitely better than what github is atm.
There are rough edges and the seeding thing is a bit mehhh. And honestly there are a bunch of things I would do differently but I like the spirit of things.
Not sure where the authors of the project stand, but it's fun to see them make progress.
AD: We're definitely interested in hearing about the rough edges and opinions! Feel free to post on our Zulip [^1] if you haven't already :).
[^1]: https://radicle.zulipchat.com/
What are the things you would do differently?
Prioritize discoverability, and self hosting on my own domain, with more customization for the protocol and ui/ux aspects.
Basically allow people to deploy something closer to gitea on a webserver initially, which has all the basic protocol features and sharing aspects implemented.
Currently the focus seems to have been very much on just the seeding & p2p part which honestly is good but not very useful for me as a developer/user who wants to work on a collaborative project.
Prioritizing P2P makes sense, but the UX/DX of radicle is not intuitive to put it mildly.
I went to the explorer radicle.network and my own as well and I still had no idea how to create an issue or submit a patch or login to an account or anything of that sort.
I opened the cli and figured out how to do all that, but then I created another node on my web server and now I have no idea how to connect to it, then I figured out how to do that, now how do I share my node without sharing it publicly, figured that out.
Cool now we are at the starting point, so every time I need to work on a project i will have to clone it, but why can't I create stuff in webui.
Maybe there is a setting I can't seem to find it yet.
But the jumping through hoops is mindbending. For a solution that could replace github for me and my friends who maintain a few public projects that wants small contributions from large number of folks as we try and maintain the local "tech" clubs.
Getting low information folks to use within 10 mins, is the bar that I have set for any platform I can use to replace github.
Radicle isn't that yet! anyways.
But I like the direction, putting contributors, access list, issues and prs in the repo is a brilliant idea, maybe put agent plans in it too, agent sessions as well maybe...
I would go as far as to say, something like a radicle network is better than every other alternative, but the UX is just not there yet.
This is all good feedback. (I'm not a radicle dev btw, just happened to be evaluating it for my own project this week.)
Fwiw, I think this is a matter of UI/UX. I think radicle provides the foundation for everything you describe -- it is just the pr/patch & issue tracking substrate, and the p2p layer for federation.
To make an analogy, I think what you are saying is "Look, BitTorrent is great and all, but I want to be able to search for music and movies. I want IMDB with a download button." That's fair, but the problem BitTorrent is solving is more fundamental, and you can build Napster or ThePirateBay on top.
Radicle, like BitTorrent, is solving the transport layer.
One challenge with distributed systems is that deletion and rollback expectations differ from traditional centralized platforms. Even if a platform removes data, cached or replicated versions can still persist elsewhere.
I'd like to see radicle replace crates.io. I can't get over Rust's dependency on github/Microsoft, and I can't get over the lack of namespacing.
All you would need is cargo compatibility, and a trusted namespace that kept up with the metadata of the current contents of crates.io, right?
edit: I really, really like rust, and love basically all of their choices about the language, but I can't stand the feeling that I'm being tricked into an ecosystem dependent on one of the worst behaved companies in the world, and I can't stand that a lot of rust projects smell like GPL-washing.
That being said, git is GPL and radicle is MIT, so it feels like the same thing, but Github also ain't git. I prefer MIT to MS; if radicle gets important enough and decides to rubpull, there will inevitably be a Free fork anyway.
* crates.io is moving away from GitHub-only authentication
* crates.io's attachment to GitHub is a fact about crates.io specifically, not the Cargo crate registry protocol
* Cargo's support for Git repositories is generic across Git and has nothing to do with GitHub specifically
* Radicle offers nothing to a crate registry that a Git remote doesn't
* and none of this has anything to do with the GPL.
It feels like you're just listing off things you like and don't like aesthetically. They have nothing to do with each other structurally.
> crates.io is moving away from GitHub-only authentication
Crates.io has not moved away from Github-only authentication, and got into the habit of yelling at people who complained about it.
> crates.io's attachment to GitHub is a fact about crates.io specifically, not the Cargo crate registry protocol
Is this just trivia you wanted to share? I feel like I covered it in the second sentence of the comment you're replying to.
> Cargo's support for Git repositories is generic across Git and has nothing to do with GitHub specifically
I'm looking to compile Rust projects from the semi-standard commonly-used crates. I do not want a Github account.
> Radicle offers nothing to a crate registry that a Git remote doesn't
Radicle offers peer-to-peer hosting that does not require a Github account.
> none of this has anything to do with the GPL.
Radicle is a project being built in Rust that partially reimplements git. Git is GPL-licensed, Radicle is MIT-licensed.
> It feels like you're just listing off things you like and don't like aesthetically.
I am unconcerned about your feelings. What I was saying is that I would like a peer-to-peer hosted, namespaced code repository that mirrors (or replaces) crates.io, and I do not want a github account to be necessary to use it.
> They have nothing to do with each other structurally.
I have no idea what "they" is referring to in this sentence.
Does radicle have some way of storing binaries outside of the source tree? I know cargo compiles from source by default, but AFAIK it can (and does) download binaries as well.
AD: Theres a project `artifact-cob` [^1] that enables some of the functionality you are looking for. You can host artifacts off-network and make 'releases' that point to them.
[^1]: https://radicle.zulipchat.com/#narrow/channel/369274-General...
radicle-artifact is currently under active development: https://radicle.network/nodes/iris.radicle.network/rad:z4VYy...
How well does it hold up under load? What are the CI and PR stories?
Their CI take is early but promising: https://radicle-ci.liw.fi/
This just describes a watcher service that kicks jobs off on an external CI system and logs the results? Not much more detail than that.
Gitlab and Github have pages and pages going over the domain language used to configure the job triggers. Jobs can trigger other jobs either in response to completion or as a dependency etc etc.
I would say these radicle-ci designs as they are now are actually quite rudimentary. That's perfectly fine for an early project but at this point I think you have to say that they won't have a CI system ready for quite some time.
I don't understand why you'd want them to built another CI system. Just for the sake of it?
Feels like using a free Jenkins gets you everything you want. Doesn't need more than a container or Java either.
That's also perfectly fine if that's what it is. I would just phrase that as "They don't have CI but they provide hooks."
it's a little of both: on the one hand we're working on CI integrations (through generic webhooks or CI-engine-specific adapters that basically implement the CI engine's API [1]), so you can keep on using your existing CI solution. On the other hand, yes, there is also work towards a new CI engine (Ambient [2]), which aims to make it safe and secure to run CI on other people's code, which is important when working towards a distributed, community-ran, CI system. We should have more docs regarding the CI story up on radicle.dev in the next week or two.
[1] - https://radicle.network/nodes/index.radicle.garden/rad%3Az3G... , [2] - https://ambient.liw.fi/
I like this idea a lot! I need to find people to try it with, though, which is not so easy with GH being so popular :-(
Some nitpicks:
* What is with the forced serif font on the website?
* Does this support other version control systems? Like mercurial, SVN, pijul, etc.?
AD: The website is currently being redesigned, I suspect we'll have it up in the next few weeks.
No it doesn't currently support other VCS's but we have planned for that possibility in future!
VCS agnosticism would be a selling point.
How would that work? Radicle is deeply tied to git. It stores all its info as git objects.
That's why architecture matters.
Would it? Realistically, what support could they add that would make it a selling point? Mercurial?
Jujutsu + stacked PRs.
It does work with jujutsu. Not sure about stacked PRs.
https://radicle.dev/2025/08/14/jujutsu-with-radicle
Does that mean it works with other compatibility layers like hg-git?
I read that it works well with jujutsu
yup ! https://radicle.dev/2025/08/14/jujutsu-with-radicle
I tried to understand what this does...
> What is Radicle? How is it different from Git/GitHub?
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git. Unlike centralized platforms like GitHub, there is no single entity controlling the network or user data. Repositories are replicated across peers in a decentralized manner. Radicle is an alternative for people and organizations who want full control of their data and user experience, without compromising on the social aspects of collaboration platforms.
(Quote from their FAQ).
This isn't even trying to answer the titular question... None of them, actually.
So, what is Radicle? A platform built on Git? What does this mean? A platform for what? What is it for?
Why Git/GitHub are used as if they were the same category of things? There's not even an attempt at answering the "how is this different from Git?" question. What does it offer that Git doesn't? Wtf is "forge"?
Radicle is an alternative... to what? I believe I have full control of my data in my Git repository... why do I need an alternative with even more control? How will I have even more control?
* * *
Maybe whatever this software does is actually useful or even good, but the documentation can't be worse.
I agree it's not immediately clear how it works, although I think I understand the role it's intending to fill.
If you're not familiar with the distinction between git and github it could be even more confusing.
As soon as I hear decentralized I have lots of questions about the underlying protocols. Their protocol page helps a little but also uses terms I'm not familiar with like "gossip protocol".
It would be nice for there to be a page that motivates the project a bit more, ie. explaining the technical problems they are attempting to solve before enumerating the components of the complex system they've built.
Someone looked at git - a distributed version control system that already works, that has been working since 2005, that Linus Torvalds wrote in a fit of pique and spite, and that currently hosts approximately all of the world's source code - and said, "This is good, but what if we added BitTorrent?" And then, presumably after consuming substances that I cannot legally inquire about, they continued: "And what if we added the Bitcoin peer-to-peer protocol?" And then, reaching a crescendo of architectural ambition that would make Icarus say "maybe dial it back a little," they concluded: "And we should DEFINITELY add blockchain identity."
Notice how each additional sentence makes the previous sentence worse, like a turducken of solutions looking for problems, or a nesting doll where every layer is a different kind of sadness. This, I presume, is what happens when you have a hammer and a screwdriver and a chainsaw, and you decide that every problem would be better solved by using all three simultaneously while riding a unicycle.
But seeing as it already does exist, it's pretty awesome.
AD: You missed the part about CRDT's!
I couldn’t find anything sarcastic to say about them. CRDTs are awesome :)
> So, what is Radicle?
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git.
-----
> Why Git/GitHub are used as if they were the same category of things?
They are not. Github is a centralized collaboration platform built on git, and radicle is a peer-to-peer collaboration platform built on git.
-----
> Wtf is "forge"?
A word some people started using for the class of Github/Bitbucket(RIP) or even Fossil-type things, as FOSS alternatives began to multiply.
-----
> Radicle is an alternative... to what?
To Github, or other "forges."
The first I remember using it was SourceForge (1999).
> To Github, or other "forges."
Then why do they bring Git into the picture? They are not comparing themselves to Git...
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git.
This is a word salad that means nothing... more than 99% of moderns software is built on Git in one way or another. Anything that is designed to be used by more than a single user could be arguably called a "collaboration platform". This description completely fails to describe anything useful about the program they are trying to describe. For instance, Git is a peer-to-peer code collaboration platform built on Git. And the same can be said about a huge number of programs that share very little in terms of purpose or application.
When someone writes a definition, s.a. you'd find in encyclopedia, the rule is that it has to (a) link to the broader category of things (b) specify in what way the subject is (mostly) unique in the category (a). If you give only (a), then the reader walks away wondering how is subject different from anything else in (a). If you only give (b), then the reader needs to guess (a), and if they fail, they may misattribute or simply abandon efforts to understand the subject.
This attempt at "definition" is the textbook example of forgetting the (b). It's something that a 10-12 y.o. could come up with... this is not what an adult should strive for.
> They are not. Github is a centralized collaboration platform built on git, and radicle is a peer-to-peer collaboration platform built on git.
You didn't understand the question.
Built on meaning the technology is using Git under the hood, not that it is developed using git.
Edit:
Breaking down the “word salad”:
> Radicle is a peer-to-peer code collaboration platform (“forge”) built on Git.
Peer-to-peer: it functions with individual nodes on the network spreading state for tracking it without relying on a single entity or centralised service.
Code collaboration platform (forge): you use it not just to store code but provides a way to keep track of “patches” (their term for PRs) and issues, amongst other things, to enable multiple people to collaborate on a code base
Built on git: the technology runs on top of git insofar as not only is the VCS just git, but the issues, patches, etc are stored in git. So the project isn’t merely developed using git, but when running the tool yourself it’s still backing everything under git.
I did understand the question. I think you're somehow confusing "built on git" with "we used git while we built it." But that's honestly just a weird guess. I really can't figure out how you could be confused about any of this.
I hope the definitions shared in the sibling comment will help.
> Then why do they bring Git into the picture? They are not comparing themselves to Git...
I would find it very strange if Github didn't mention git. It would be equally strange if radicle didn't mention git. They both wrap git.