Writing the master’s thesis in French as a Foreign Language (FFL) is a complex academic activity that requires constructing a research object, problematizing, mobilizing knowledge, ensuring methodological coherence, and building a scientific argument. Yet the assessment criteria commonly used often remain insufficiently operational, rarely allowing the writer to pinpoint the gaps in a text or to undertake the revisions needed to close them. This article sets out to design a formative assessment device capable of converting the scientific requirements of the thesis into resources that can be mobilized for diagnosis and revision. The study draws on a diagnostic analysis of 25 master’s theses in FFL didactics. Combining academic literacies, genre theory, and formative assessment, the analysis organizes criteria along a functional chain linking the scientific requirement, the intellectual operation, its discursive realization, an observable indicator, textual evidence, diagnosis, and revision. The results chiefly reveal difficulties with scientific positioning, the synthesis of prior work, its critical confrontation, the interpretation of results, and self-regulated revision. They also show that the formal presence of a thesis component does not guarantee its effective scientific realization. The study accordingly proposes treating the assessment criterion not as a mere instrument of judgment but as an operational unit linking observation, diagnosis, and revision. While the analysis establishes the diagnostic relevance of the proposed architecture, its formative effectiveness still needs to be tested in authentic writing and revision situations with students.