Safewards is implemented internationally to prevent conflict and containment in psychiatric settings. Although evaluations generally show a reduction in such incidents, effects vary across studies. A key limitation is the absence of a valid and robust measure of implementation fidelity to indicate how closely Safewards is implemented as originally intended. Without this knowledge, replication, refinement, and effective implementation is difficult, jeopardizing both clinical quality assurance and scientific evaluation rigour. This study reports the first steps in developing a new implementation fidelity tool for Safewards. The objective was to identify important active ingredients and corresponding measurement levels for the 10 Safewards interventions. A Delphi survey design with an international expert panel was used to reach consensus on active ingredients and measurement levels. The preliminary survey list was derived from the Safewards.net website and informed by implementation fidelity theory, incorporating both direct and indirect measurement levels. The resulting set of ingredients and measurements was pilot tested for comprehensibility by likely end-users in an acute psychiatric clinic. Their experiences were explored in a group interview, analyzed with thematic content analysis, and informed adjustments. The final output comprises 48 active ingredients and 122 measurements across the 10 interventions, each represented by three to seven statements and up to four measurement levels. The findings provide a conceptual and content-valid foundation for the development of a new Safewards fidelity tool. In addition, the active ingredients have intrinsic value, offering guidance for the clinical implementation of Safewards.