Sources: Mark Zuckerberg is back to writing code after a two-decade hiatus, submitting three diffs to Meta's monorepo, and is a heavy user of Claude Code CLI
anthropic claude meta
| Source: Mastodon | Original article
Mark Zuckerberg has resurfaced as a hands‑on programmer, submitting three code diffs to Meta’s sprawling monorepo for the first time in twenty years. The changes, modest in scope but notable for their provenance, were pushed through the company’s internal review system on April 4, 2026. Each diff was generated with the Claude Code command‑line interface, Anthropic’s AI‑assisted coding tool, which Zuckerberg has reportedly been using daily to prototype and debug.
The episode matters on several fronts. First, it signals a rare public re‑engagement of the CEO with the technical core of his own platform, a move that could reshape internal culture by reinforcing a “code‑first” ethos among senior leadership. Second, Zuckerberg’s reliance on Claude Code highlights the growing penetration of generative‑AI assistants in enterprise software development, especially at scale. Meta’s monorepo, which houses everything from the News Feed to the Reality Labs stack, is notoriously complex; an AI‑driven workflow that can produce production‑ready patches may accelerate feature rollout and reduce bottlenecks.
Anthropic’s simultaneous announcement that, effective 12 p.m. PT on the same day, Claude subscriptions will no longer cover usage on third‑party platforms adds a strategic layer. If Meta’s internal tooling depends on Claude Code, the policy shift could force renegotiations or spur Meta to develop its own AI‑coding assistant, intensifying the AI‑tool arms race among tech giants.
What to watch next: whether Zuckerberg continues to commit code and if his patches influence larger product decisions; how Meta’s engineering teams adapt to the new Anthropic licensing terms; and whether other executives follow suit, turning AI‑augmented coding from a niche experiment into a mainstream practice across the company. The convergence of leadership, AI tooling, and licensing policy could reshape how the next generation of software is built at Meta and beyond.
Sources
Back to AIPULSEN