In this project, I argue that agency is an important soft skill for improving the life and labor market outcomes of vulnerable adolescent girls, and so should be a priority in international education policy and for achieving quality education for all (goal 4 in the United Nations SDG agenda). I argue that the way agency is defined, measured, and practiced should attend to the contextually-specific ways in which agency is constrained, and to the power relationships that have historically privileged international actors over local practitioners, and methodological concerns about internal validity over external validity. These dynamics in the girls’ education sector have prevented the emergence of accurate measures and truly scalable approaches to programming. I argue that the practice and measurement agency should be ‘emplaced’ or situated within the particular economic, cultural, and socio-political milieu that structures girls’ lives and opportunities, and I provide examples of how to do this. In paper 1 I use a review of the education, development, and psychology literature, a benchmarking of girls’ life skills curricula, and two empirical case studies to synthesize a working definition of agency and to propose a theory for how agentic capacity is developed through skills-focused or critical pedagogical approaches to programming. In paper 2, I offer an example of how to address the problem of contextually relevant measurement, by developing a measure for adolescent girls’ agency in East Africa. The paper utilizes a participatory survey development process to create an ‘emplaced’ measure of girls’ agency as a 4-dimensional construct made up of both internally and externally facing skills and beliefs. I then conduct an investigation of the survey’s construct validity and validity of intended use as a pre-post program evaluation measure—using Confirmatory Factor Analysis (CFA) via structural equation modeling (SEM). In paper 3 I argue that instead of seeking to scale evidence-based interventions, the sector should identify scalable theories of practice. I use Group Concept Mapping as a participatory tool for group conceptualization and theory-building, to construct a theory of practice for replicating adolescent girls’ agency outcomes––one that is empirically driven and collaboratively built by the expertise of local practitioners and community-driven organizations (CDOs) in East Africa.