Cosesaurus is a genus of archosauromorph reptiles likely belonging to the family Tanystropheidae. It is known from fossil imprints of a single small skeleton, MGB V1, which was found in Muschelkalk outcrops near the municipalities of Mont-ral and Alcover in Spain. These outcrops are dated to the Ladinian age of the middle Triassic about 242 to 237 million years ago. The specimen is stored at the Museu Martorell (a.k.a. the Museu Geologia de Barcelona), which is now part of the Museu de Ciències Naturals de Barcelona. The poor preservation and likely juvenile nature of the specimen has led to the anatomy of Cosesaurus being misidentified by several different sources. For example, Paul Ellenberger claimed that it was an ancestor to birds in the 1970s, while David Peters claimed that it was a pterosaur ancestor in 2000. Both of these claims contrast with mainstream scientific theories on the origins of either group, and other paleontologists who study the specimen are unable to find the features which Ellenberger or Peters reported to be present. The Ellenberger and Peters hypotheses are thus considered fringe theories with questionable scientific soundness due to their low reproducibility. Mainstream hypotheses for the relations of Cosesaurus generally agree that it is a "protorosaur", specifically a tanystropheid closely related to long-necked reptiles such as Macrocnemus, Tanytrachelos, Tanystropheus, or Langobardisaurus.

Due to its poor preservation in grainy sedimentary rock, the fossil has been misinterpreted by many paleontologists in the past. The first describer of the specimen, Paul Ellenberger, erroneously observed bird-like features which could not be confirmed by later studies. He published several analyses of the specimen, in 1974, 1977, and 1993. He hypothesized that it was an ancestor to birds, but this hypothesis was soon abandoned by other paleontologists.

For example, in 1979 Paul Olson briefly discussed the specimen during his description of the small tanystropheid Tanytrachelos. He found difficulty diagnosing the specimen using Ellenberger's photographs, but he did note that it shared many similarities with tanystropheids such as Tanytrachelos and Tanystropheus rather than birds.

Ellenberger's hypothesis fell out of favor in the late 20th century due to the growing evidence for the origin of birds among coelurosaurian dinosaurs. During this time, most comprehensive study of Cosesaurus since Ellenberger's was performed by José Luis Sanz & Nieves López-Martínez in 1984. They found that barely any of the features Ellenberger used to link Cosesaurus to birds were valid. They also studied its proportions and concluded that it was likely a juvenile specimen. Based on this hypothesis, they constructed a skeletal diagram showing what an adult Cosesaurus may look like, considering how the proportions of reptiles change during development. Sanz and López-Martínez found that the closest relative of Cosesaurus was probably a "prolacertid" such as Macrocnemus.

In 2000, a variant of Ellenberger's hypothesis was published by researcher Dave Peters. Peters claimed to have found features in the specimen which resemble those which Ellenberger claimed connected Cosesaurus to birds. However, Peters instead purported that these putative features indicated that Cosesaurus was an ancestor to pterosaurs rather than birds. However, paleontologists who examined the specimen could not find evidence to support Peters' claims, and noted that his link between Cosesaurus and pterosaurs relied on a poorly constructed phylogenetic analysis. The consensus view, as explained and advocated by studies such as Nesbitt (2011) and Ezcurra (2016), is that pterosaurs were not closely related to "protorosaurs" such as Cosesaurus, and instead are advanced archosaurs closely related to dinosaurs. In 2009, Peters published a follow-up to his 2000 study which used a digital tracing algorithm to support his claims for features such as prepubic bones and frill-like skin flaps. However, this method has also been heavily criticized due to its reliance on photographs in the literature rather than personal observation or CT scan evidence.

A doctorate thesis published in 2016 by Franco Saller, although mostly focusing on the musculoskeletal system of Macrocnemus bassanii, also provided a redescription of Cosesaurus. This thesis, written in Italian, debunked certain features of the specimen purported by Ellenberger or Peters, such as the presence of a sternum or more than two vertebrae in the sacrum. It also provided additional implications for the anatomy and biology of members of the genus.

The specimen has an estimated length of 14 centimeters (5.5 inches), although the impressions of the tip of the tail are faint, so this length may have been slightly longer. Due to the poor preservation of the specimen, definitive features are difficult to find. Ellenberger reported webbed feet and feather impressions imprinted among the bone impressions, and Peters (2000) claimed that Ellenberger instead observed skin remnants such as a patagium or actin fibers. However, both of these claims of soft-tissue are dubious, as other paleontologists who have analyzed the specimen explain that the texture of the rock surface near the bones cannot be differentiated from the rest of the rock slab.

The snout is moderate in length, not as elongated as the state in early theropod dinosaurs like Coelophysis, but not as short as in early amniotes. Few details of the skull can be identified with absolute confidence. The skull and particularly the eyes were proportionally quite large, although these are likely juvenile features. The dome-like skull roof was thinnest above the eyes before broadening towards the back of the head. Numerous small, pointed teeth were present in the mouth, including an estimated six or seven in the long and pointed premaxilla. A few teeth at the back of the mouth were large, triangular, and blade-like compared to the thinner teeth in the rest of the jaw. Although the skull impression is crushed, a small depression in front of the eye is conspicuous. Ellenberger (1977) claimed that this depression was perforated through the snout and represented an antorbital fenestra, while Peters (2000) argued that it represented a series of fenestrae. Close examination by Saller (2016), however, showed that the depression's depth could not be determined, nor which bones surrounded it. The jaw joint possesses some features which indicate that Cosesaurus was a saurian, similar to modern reptiles such as lizards, dinosaurs, and crocodilians, instead of more primitive amniote groups such as araeoscelidians and weigeltisaurids. These features include the possession of a retroarticular process of the mandible and a quadrate bone which is notched from behind and visible from outside the skull. However, the flexibility of the quadrate (and therefore the potential for cranial kinesis) is unclear.