In recent years, global avian diversity has been in continuous decline. Traditional monitoring methods are often constrained by detectability limitations and observational challenges, which makes it difficult to fully characterize community structures. Environmental DNA (eDNA) metabarcoding offers a promising alternative for avian diversity monitoring. However, universal eDNA primers for birds remain scarce, and existing ones often lack sufficient amplification universality and taxonomic resolution for reliable application in complex environments. To address these limitations, this study developed a new set of universal avian eDNA primers, named LiBird, and systematically evaluated their performance under multiple conditions. In silico PCR analysis showed that LiBird achieved a 99.3% amplification success rate against a reference database of avian mitochondrial genomes, outperforming the widely used MiBird primers (88.3%), with significantly fewer primer-template mismatches. In a mock DNA community comprising 30 bird species, LiBird detected all species, whereas MiBird detected 28. In controlled zoo water samples, LiBird detected 26 bird species, compared to 19 by MiBird, demonstrating its higher amplification efficiency and detection capacity. Field validation was conducted using natural water samples from the Chifenggang Wetlands, Nanhui New City, Shanghai, where the LiBird primers successfully detected 38 bird species, including several species that are difficult to spot through visual observation, such as the Baillon's Crake (Zapornia pusilla). The results indicate that LiBird performs well in terms of amplification, species coverage, and detection sensitivity, making it a reliable tool for avian eDNA monitoring. Nonetheless, the study revealed that a single primer system still carries a certain risk of missing species; for instance, in environmental samples, both MiBird and LiBird were able to detect some species that the other primer set missed. Thus, combining multiple primers and diversifying environmental sample types are recommended to improve detection comprehensiveness and data reliability in future avian biodiversity surveys.