Generative AI is transforming how software is created and consumed. A growing trend known as "vibe coding" involves AI agents autonomously building software by selecting and assembling existing open-source software (OSS), often without users directly engaging with documentation, contributing fixes, or reporting bugs. This shift raises important questions about the long-term sustainability of the open-source ecosystem. This paper examines the equilibrium effects of vibe coding on the open-source landscape. It presents a model featuring endogenous project entry and varying levels of project quality, where OSS serves as a scalable input for producing new software. Users face a choice: they can either use OSS directly—engaging with its code, documentation, and community—or rely on AI agents to integrate it through vibe coding. While vibe coding significantly increases productivity by reducing the cognitive and technical barriers to using and building on existing code, it simultaneously weakens the feedback loops that sustain open-source development. In many cases, maintainers depend on user engagement—such as contributions, bug reports, and community support—for recognition and indirect rewards. When this engagement declines due to reliance on AI agents, the incentives for developers to create and maintain high-quality open-source projects diminish. The study finds that when OSS is monetized solely through direct user engagement, widespread adoption of vibe coding leads to lower entry into the ecosystem, reduced sharing of code, and a decline in both the availability and quality of open-source software. Paradoxically, even though overall productivity rises, total welfare falls due to the erosion of the open-source commons. To preserve the current scale and quality of open-source software under widespread vibe coding, substantial changes in how maintainers are compensated are necessary. This may include new funding models such as direct grants, corporate sponsorships, or AI-driven revenue-sharing mechanisms that ensure contributors are rewarded even when end users do not interact directly with the code. The findings underscore a critical challenge: while AI-driven development boosts efficiency, it risks undermining the collaborative foundations of open-source software. Sustaining this vital resource in the age of generative AI will require rethinking how value is captured and distributed within the ecosystem.