Tangled Launches First Truly Federated Code Forge, Shifting Away From Centralized Platforms

Tangled announced the release of what it calls the first truly federated code forge, a platform that lets developers host Git repositories, CI runners and the UI on their own infrastructure while still collaborating with anyone on the network. The move challenges the dominance of centralized services like GitHub and GitLab, promising greater control over uptime, data stewardship and extensibility.

What makes Tangled’s forge "federated"?

According to the Tangled Blog, the new forge is built on the open AT Protocol, which separates the transport layer (email‑style messaging) from the code storage layer (Git). Users can run only the components they need—such as a Git server, a CI runner, or the full API/UI stack—and still interoperate with other hosts. The architecture stores issues, pull requests, comments and other collaborative objects (COBs) directly inside the Git repo as signed records, making the full project portable across any Tangled instance.

How the platform’s architecture has changed

The previous version relied on a monolithic Go service (the appview) that combined AT Protocol indexing, caching and UI rendering on a single SQLite database in Sweden. The new release replaces that with a modern stack:

• Bobbin – a disk‑less, in‑memory indexer written in Rust, deployed to more than 11 edge regions for low‑latency reads.
• Knot2 – a Rust‑based Git and SSH server that also stores COBs alongside code, turning each repository into a self‑contained data store similar to a Personal Data Server (PDS).
• Svelte UI – a fresh front‑end built on top of the edge‑deployed Bobbin, offering faster, visually consistent interactions.

These components communicate via the AT Protocol firehose, ensuring that updates to code or metadata are instantly synced across the federation.

New developer‑focused features

Tangled introduced several workflow enhancements aimed at modern development practices:

• Organization accounts – Real AT Protocol accounts with DIDs and password‑less delegation, allowing owners to invite members and act on behalf of the org when creating or forking repos.
• Stacked PRs – A revamped pull‑request flow that supports stacked changes for both vanilla Git users and Jujutsu users, including inter‑diff views between versions.
• Upcoming private repos, CLI, SDKs and agent skills – The team plans to roll out these capabilities over the coming weeks, extending the platform’s appeal to enterprise and automation use‑cases.

Why self‑hosting matters for developers and researchers

Self‑hosting gives teams full control over the runtime environment, compliance requirements and cost structure. Because issues, PRs and other metadata travel with the repo, projects can be moved between servers without losing history or needing migration scripts. The attestation model—where every record written to a knot also creates a signed attestation in the author’s PDS—provides two‑way verification that mitigates the risk of unauthorized changes.

For researchers who need reproducible environments, the ability to run a custom knot on a platform like Cloudflare Workers (as demonstrated by Tangled’s own Artifacts‑backed knot) means experiments can be packaged with both code and its full discussion history.

How Grafy users can explore Tangled’s concepts

While Grafy does not host a federated forge itself, creators and developers can experiment with the ideas using existing Grafy tools:

• Grafy Code – Use the AI‑assisted coding studio at /code to prototype a simple Git‑backed project, then export the repo and run a local Tangled knot for hands‑on testing.
• Grafy Reader – Import the Tangled documentation PDF (or any technical spec) into /reader to ask questions about the AT Protocol, COBs or upgrade steps, helping you understand the protocol without leaving the workspace.
• Grafy Watch – Set up a monitor on a public Tangled repository URL with /watch to receive AI‑summarized updates whenever a new issue or PR lands, mimicking the federated notification flow.

These workflows let you experience the benefits of decentralized collaboration while staying within the Grafy ecosystem.

What’s next for Tangled and the broader ecosystem?

The release is being staged over several weeks, with private repositories, a full CLI, language SDKs and agent‑skill extensions slated for upcoming sprints. As more developers adopt the AT Protocol‑based model, we can expect a growing ecosystem of interoperable tools—potentially including Grafy‑native integrations that could directly pull data from a Tangled knot for AI‑driven code reviews or documentation generation.

---

Source: Tangled Blog – "The federated code forge" (30 October 2026) – https://tangled.org/blog/forge/

Frequently asked questions

How does Tangled ensure that issues and pull requests stay in sync with the code repository?

Tangled stores issues, PRs, comments and other collaborative objects as COBs—Git objects embedded in the repository itself. Each change is signed and also recorded as an attestation in the author’s PDS, providing cryptographic proof of authenticity and enabling automatic sync via the AT Protocol firehose.

Can existing GitHub projects migrate to Tangled without losing history?

Yes. Because Tangled treats the repository as a self‑contained data store, you can push the full Git history (including COBs) to a Tangled knot. The platform’s upgrade guide (v1.20.0) details the steps for moving a repository and its metadata to the new architecture.

What role do the new "bobbin" and "knot2" services play in the federated model?

Bobbin is a read‑only, in‑memory indexer that serves AT Protocol data from edge locations, delivering fast UI responses worldwide. Knot2 is the core Git/SSH server that also writes collaborative objects to the repo, effectively turning each repository into a mini‑data server that participates in the federation.

---

Source: Hacker News — front page — The Federated Code Forge