This article is written in Portuguese to address the significant number of Portuguese-speaking developers interested in the Universal Trust Adapter (UTA). Previously, information was only available in English. The core problem UTA solves is the lack of a standardized way for AI agents to verify the credentials of tools they invoke. Currently, different standards answer questions about credential issuance, validity, and scope inconsistently, leaving AI agents confused. UTA acts as a universal verifier supporting eight different credential formats, including JWT, W3C Verifiable Credentials, and X.509 certificates. It processes these credentials through a detailed 12-step verification pipeline. Developers can use UTA via a public API or an NPM package for easy integration into their AI agent projects. The project boasts impressive real-world benchmarks, demonstrating high performance and extensive testing. UTA is particularly relevant for Brazil's large developer community, especially in sectors like fintech and healthtech where agent trust is paramount. It's designed to be small, fast, and dependency-light for easy integration. However, UTA does not address honesty or reputation layers, nor does it provide pre-defined policies; these require separate implementations. The author encourages developers to read the README, test the API, and contribute to the project through issues. Ultimately, UTA aims to bridge existing trust standards rather than create a new one, facilitating interoperability for AI agents.