Scholarly publications today are still mostly disconnected from the underlying data and code used to produce the published results and findings, despite an increasing recognition of the need to share all aspects of the research process. As data become more open and transportable, a second layer of research output has emerged, linking research publications to the associated data, possibly along with its provenance. This trend is rapidly followed by a new third layer: communicating the process of inquiry itself by sharing a complete computational narrative that links method descriptions with executable code and data, thereby introducing a new era of reproducible science and accelerated knowledge discovery. In the Whole Tale (WT) project, all of these components are linked and accessible from scholarly publications. The third layer is broad, encompassing numerous research communities through science pathways (e.g., in astronomy, life and earth sciences, materials science, social science), and deep, using interconnected cyberinfrastructure pathways and shared technologies. The goal of this project is to strengthen the second layer of research output, and to build a robust third layer that integrates all parts of the story, conveying the holistic experience of reproducible scientific inquiry by (1) exposing existing cyberinfrastructure through popular frontends, e.g., digital notebooks (IPython, Jupyter), traditional scripting environments, and workflow systems; (2) developing the necessary 'software glue' for seamless access to different backend capabilities, including from DataNet federations and Data Infrastructure Building Blocks (DIBBs) projects; and (3) enhancing the complete data-to-publication lifecycle by empowering scientists to create computational narratives in their usual programming environments, enhanced with new capabilities from the underlying cyberinfrastructure (e.g., identity management, advanced data access and provenance APIs, and Digital Object Identifier-based data publications). The technologies and interfaces will be developed and stress-tested using a diverse set of data types, technical frameworks, and early adopters across a range of science domains.