Graphite is joining Cursor

Original Article ↗ Hacker News Discussion ↗

Overall reaction to the acquisition

  • Many commenters express excitement about two “favorite” tools combining and praise the communication style and culture-focused tone.
  • A substantial contingent is deeply skeptical, citing a long history of acquisitions leading to shutdowns (“our incredible journey” trope).
  • Cursor’s prior acquisition and sunsetting of Supermaven is repeatedly referenced as a cautionary example; assurances that “Graphite isn’t going anywhere” are widely treated as non-credible because control now sits with the acquirer, not the founders.

Concerns about Graphite’s future

  • Power users of Graphite’s core features (stacked PRs, CLI, review UI) worry this is “normally not good news” and anticipate needing to migrate away preemptively to avoid a later scramble.
  • Graphite cofounders repeatedly say the product will be maintained, improved, and integrated, with more resources than before; critics respond that this has been said in many past acquisitions that still ended in shutdowns.
  • Some see this as “final nail” in the coffin of the original non-AI Graphite, recommending jj, git-spice, git-branchless, tangled.sh as alternatives.

What Graphite actually does

  • Several explanations clarify Graphite is primarily about managing stacks of dependent PRs and a better PR/review UI, not just AI review.
  • Stacked PR support (auto-rebasing, keeping branches in sync, making many small PRs manageable) is described as a “game changer” versus raw GitHub.
  • Others say they don’t see much value beyond what GitHub + a few aliases could already do.
  • There’s discussion of GitHub’s upcoming native stacked PRs; some think this could eventually make Graphite obsolete, others are unsure about GitHub’s execution or UX.

AI code review and workflow integration

  • Multiple comments note Graphite predates AI review; its AI is seen as “subpar but everything else is really good” by some.
  • Users compare Graphite/Cursor’s reviewers with tools like Qodo, Sentry, Codex, Claude Code; consensus is that diff-only AI review is limited and effective tools must index the full repo and docs.
  • Some report strong real-world value from AI review (catching non-obvious bugs); others say AI review often adds noise and can’t reliably understand business logic.

Cursor, IDE vs CLI, and strategy

  • Large subthread debates IDE-integrated agents (Cursor, Claude plugins, Windsurf) vs CLI/terminal-centric tools (Codex CLI, Claude Code), with strong opinions both ways.
  • Supporters of Cursor emphasize tight integration, vertical product polish, model-agnosticism, and enterprise controls; detractors question the moat of a VS Code fork and complain about pricing and usage-based billing.
  • On why Cursor acquires instead of “just building with AI,” several point out Graphite’s complexity, production hardening, user base, and distribution, arguing AI does not make recreating such a product trivial.