Agreed definitions of safety	Shared definitions of what constitutes a patient safety problem in this context must be developed, to agree on an agenda for improving safety.
A wider remit for safety	Evaluation of ‘safe’ community services must not be centred around a limited number of recognized adverse incidents only (e.g., incidents of violence and aggression). Attention must be paid to what makes service users feel safe or unsafe, along with broader, upstream determinants of safety (e.g., safe waiting times).
Measure safety over time	Efforts to investigate and measure safety in community‐based mental health care must be designed with long‐term care journeys in mind.
Make greater use of theory and evidence	Opportunities should be identified to learn from existing patient safety theory and evidence, which may have its origins within other care specialties or settings.
A nuanced approach to intervention	A wider range of safety interventions are needed to target systems factors which impact the safety of care, moving beyond a focus on direct service user and staff factors alone.
Draw on wider sources of ‘safety intelligence’	Beyond traditional academic evidence, we must also seek to understand what can be learned from existing unpublished literature and local quality improvement work. Likewise, our approach to safety improvement in community services must be shaped by and with mental health service users and carers.