Background Cholera outbreaks remain a major public-health challenge in sub-Saharan Africa, where diagnostic capacity is limited and clinical case definitions are non-specific and reply heavily on syndromic diagnosis. Rapid identification of Vibrio cholerae is critical, yet cholera-suspected diarrhoea can have multiple infectious causes not captured by targeted diagnostics. Methods We evaluated a mobile, culture-independent metagenomic sequencing workflow for on-site detection of gastrointestinal pathogens directly from faecal samples in Burundi. The offline workflow combined long-read ONT sequencing with rapid, laptop-based taxonomic and antimicrobial resistance (AMR) screening and was deployed across a health centre, a district hospital, and a refugee transit camp. The frontline and real-time results were verified using both conventional culturing and in-depth bioinformatic analyses. Results V. cholerae signals were only detected in a subset of suspected cholera cases, while many samples were dominated by alternative bacterial taxa, most frequently Escherichia coli. V. cholerae abundance correlated strongly with detection of the cholera toxin phage CTX{phi}, supporting differentiation between toxigenic signal and background exposure. AMR genes were detected across samples, providing early situational insight into resistance determinants among gastrointestinal bacteria. Conclusions Mobile, offline metagenomic sequencing enables rapid frontline characterization of gastrointestinal disease, especially cholera-suspected, in resource-limited settings and complements existing diagnostics by improving etiological resolution and outbreak response.