The Prompt-to-Paper framework uses a multi-agent system to generate scientific manuscripts grounded in verifiable literature. It replaces fabricated results with a deterministic retrieval-augmented generation pipeline and section-aware scoring. This approach targets the lack of rigor in AI-written papers. Researchers can now validate claims against real citations rather than relying on model hallucinations.