BRAT is a web-based tool for text annotation that integrates Natural Language Processing (NLP) technology to support manual curation and increase annotator productivity. It is designed for rich structured annotation of various NLP tasks and has been used in several real-world annotation projects. BRAT provides an intuitive interface, high-quality visualization of annotations, and versatile support for a wide range of annotation tasks. It allows users to search for text spans, relations, event structures, or text, and supports multiple annotation types, including entity mentions, relations, and events. BRAT also includes a comprehensive set of search functions and supports the Unicode standard, enabling annotation of text in various languages. The tool is available under an open-source license and is implemented using a client-server architecture with communication over HTTP using JSON. The server is a RESTful web service and supports both CGI and FastCGI protocols. BRAT also includes a system for checking detailed constraints on annotation semantics, ensuring that annotations are accurate and consistent. The tool has been used in several annotation projects, including the BioNLP Shared Task 2011, where it was used for the annotation of entity mentions and events. A study showed that the integration of a machine learning-based semantic class disambiguation system reduced the time spent on selecting annotation types by over 30% and overall annotation time by 15%. BRAT is freely available under the open-source MIT license from its homepage at http://brat.nlplab.org.BRAT is a web-based tool for text annotation that integrates Natural Language Processing (NLP) technology to support manual curation and increase annotator productivity. It is designed for rich structured annotation of various NLP tasks and has been used in several real-world annotation projects. BRAT provides an intuitive interface, high-quality visualization of annotations, and versatile support for a wide range of annotation tasks. It allows users to search for text spans, relations, event structures, or text, and supports multiple annotation types, including entity mentions, relations, and events. BRAT also includes a comprehensive set of search functions and supports the Unicode standard, enabling annotation of text in various languages. The tool is available under an open-source license and is implemented using a client-server architecture with communication over HTTP using JSON. The server is a RESTful web service and supports both CGI and FastCGI protocols. BRAT also includes a system for checking detailed constraints on annotation semantics, ensuring that annotations are accurate and consistent. The tool has been used in several annotation projects, including the BioNLP Shared Task 2011, where it was used for the annotation of entity mentions and events. A study showed that the integration of a machine learning-based semantic class disambiguation system reduced the time spent on selecting annotation types by over 30% and overall annotation time by 15%. BRAT is freely available under the open-source MIT license from its homepage at http://brat.nlplab.org.