Anhua dark tea, a protected geographical indication product in Hunan, China, derives its value from specific mountain micro-terroirs, including Furongshan, Gaoma Erxi, and Yuntaishan. Across these areas, micro-terroir and cultivar variations impart distinctive chemical components to the decisive raw material (locally known as Hei Mao Cha). For the authentication of these specific origins, we developed a rapid HPLC-DAD method coupled with the ATLD algorithm, enabling the quantification of caffeine and seven major flavan-3-ols within five minutes. Our method achieved satisfactory accuracy, with average recoveries of 84.73–119.88% and RMSEP values ranging from 0.28 to 4.39 μg/mL. We subsequently applied PCA and FCA, which revealed distinct clustering patterns of the tea samples by their mountain origin. Notably, Furongshan and Gaoma Erxi exhibited markedly distinct chemical profiles, while Yuntaishan showed intermediate characteristics. This integrated HPLC-DAD/ATLD protocol, coupled with non-linear t-distributed stochastic neighbor embedding (t-SNE) followed by random forest (RF) classification (validation accuracy: 85.7%), offers a practical solution for the fine-scale geographical traceability of raw dark tea, supporting quality control and providing insights into how micro-terroir shapes tea chemistry. This approach can be readily adapted for the authentication of other geographical indication products.