D2 diagramming language goes non-profit under Hack Club as its parent company shuts down

D2 moves under Hack Club's fiscal sponsorship

D2, the open-source, text-based language for describing software diagrams, is now fiscally sponsored by Hack Club, according to an announcement on d2lang.com that reached the front page of Hacker News on September 5, 2026. The move turns D2 into an independent, fully open-source, non-profit project — and it was prompted by sobering news: the for-profit company that created and bankrolled the language is shutting down.

From open-core arm to community asset

D2 began life as the open-source half of an open-core business. Work on the free language was funded by two closed-source products, the D2 Studio IDE and the proprietary TALA layout algorithm. With the company winding down, D2 continues outside that structure, and the announcement promises that both proprietary pieces will be released, with follow-up posts planned as TALA and D2 Studio become available.

The project's creator, Alex, writes that he will keep maintaining and improving D2, although with limited time because it is no longer his main focus. Interested maintainers are invited to email him, and he notes that consistently high-quality contributions could earn paid contributor contracts funded by donations to the non-profit — an arrangement he frames as well suited to a driven early-career developer with open-source experience.

An AI-driven next phase

The most unusual part of the announcement is how D2 will be built from here. The creator says all of his contributions will now be produced with AI, AI-made contributions from others are welcome, and AI will review those submissions as well. Only prose — blog posts and communications — will stay human-written.

He expects this to let the project advance many times faster than before, with models optimising performance in ways he could not and delivering features that were previously out of reach for a small team, such as an isometric renderer and transpilers from other diagramming languages. He concedes the approach is controversial and has downsides, points out that he coded almost daily for more than a decade but has written none in the past year, and says he will not let the codebase's quality degrade.

The stated priorities are:

• Native rendering support on widely used platforms such as GitHub, Notion and Linear.
• A stable 1.0 release with a formalised grammar and feature set.
• Becoming the strongest output format for AI agents, on the reasoning that agents produce diagrams that are close to correct yet still non-deterministic and imprecise, so diagrams still need a text-based source behind them — similar to how HTML sits behind a rendered page rather than every pixel being drawn by hand.

Deliberate limits

The announcement is just as clear about what D2 will not do. To avoid bloat, the language will keep a small, curated feature set with an emphasis on depth, plus a plugin or module system for anything outside the core. D2 will stay focused on software architecture diagrams; Venn diagrams are ruled out even if they would be trivial to add.

There will be no built-in LLM interface either — the creator's view is that models can already study examples and turn specs or code into .d2 files, so a bespoke chat wrapper adds nothing. And because D2 is a non-profit, anything that requires servers is off the table: no MCP, no API, no user accounts, no server-side rendering, no multiplayer. D2 Studio will run entirely client-side and work offline.

Why it matters

Teams that embedded .d2 files in their documentation faced the classic risk of a language tied to one company. Instead of a quiet sunset, D2 gets continuity: a non-profit steward, a pledge to release the paid IDE and layout engine to everyone, and a donation-funded path to paid maintenance. That is a template other open-core companies winding down could copy.

It is also a live experiment. Reduced maintainer availability, AI-first development and donation-funded contributor contracts form an untested mix that the creator himself flags as controversial. How it holds up will interest anyone watching whether AI can substitute for the engineering bandwidth that small open-source projects have always lacked.