Launch HN: Haystack (YC S24) – Visualize and edit code on an infinite canvas

Original Article ↗ Hacker News Discussion ↗

A new code editor called Haystack, built as a fork of VS Code, introduces an infinite canvas that visualizes relationships between files and symbols to reduce cognitive load when navigating large or legacy codebases. Commenters are split on whether this spatial UI is more efficient than traditional keyboard- and LSP-driven workflows, with many requesting stronger keyboard support, automated layout, and eventually a lighter-weight extension that can run inside existing editors like VS Code or Cursor. Others focus on practicalities such as licensing restrictions, enterprise adoption, and the planned business model centered on collaborative features and optional generative AI, while suggesting future directions like 3D/VR views and whole-system visualizations across services.

Product concept & value proposition

  • Haystack is a VS Code fork that puts code editors on an infinite 2D canvas, auto-linking related files/symbols as you navigate.
  • Goal: reduce cognitive load when understanding large/legacy codebases, support spatial reasoning, and make multi-file flows easier to grasp.
  • Several devs relate it to their own ad‑hoc pane/tab setups, method-of-loci style memory, and visual tools like Miro/Heptabase.

VS Code fork vs. extension

  • Many commenters want this as a VS Code extension, or compatible with Cursor, to avoid switching IDEs and easing enterprise adoption.
  • Authors say extensions can’t deeply change UI or treat multiple embedded editors as first-class tabs, hence the fork; a “lite” extension is planned.
  • Some see the proliferation of VS Code forks as risky and think platform ownership/monetization is the real driver; others cite Cursor as a counterexample that a fork can succeed.

UX, keyboard use, and monitor constraints

  • Enthusiasts like the idea of spatial layout, especially with ultrawide monitors or VR/3D concepts.
  • Skeptics find canvas UIs slower than LSP + keyboard workflows and worry about mouse dependence and clutter with many editors.
  • There are keybindings and some tiling/auto-positioning; users request better demos of pure-keyboard workflows and more automation in layout.
  • Vim plugin mostly works but has some navigation bugs. Laptop users worry about limited screen real estate.

Visualization depth & features

  • Requests for:
    • Caller/callee graphs and semantic zooming.
    • 3D or layered visualizations, clustering, and background/minimap aids.
    • Sticky notes, diagramming, documentation mixing, UML-like overviews, multi-service graphs, and visual diff/test-coverage views.
    • Cross-repo/API/service linking, possibly via AI.

Licensing & enterprise concerns

  • Current Polyform Strict-style license is seen as too restrictive and confusing for commercial use and contributions; authors plan to adjust it.
  • A few note they can’t adopt it at work until the license risk is clarified.

Monetization & roadmap

  • Planned revenue from collaborative and generative-AI features: shared workspaces, natural-language search, multi-file edits.
  • Some are enthusiastic about collaboration; others are wary of “GenAI” focus.

Technical scope

  • Works with VS Code extensions and aims to track upstream.
  • Python support is good; C# is limited by ecosystem features, especially for call graph details.