DataQuoll

Most tools have clearly distinct purposes, especially in the declarations, gauges, and calendars groups. The incident/event family (list_incidents, list_events, get_incident, get_event, incident_snapshot, nearby_incidents) is the only real source of ambiguity, but the descriptions clarify the clustered-event versus raw-incident distinction.

The set mostly follows a list_/get_ verb-noun pattern, but it mixes in noun-phrase names like incident_snapshot and nearby_incidents, and uses inconsistent singular/plural in declaration_by_agrn versus declarations_by_point/declarations_by_postcode. Still readable and predictable overall.

22 tools is on the heavy side for an MCP server, with several closely related variants for incidents, declarations, and gauges. However, the server spans multiple coherent subdomains—incidents, floods, declarations, and calendars—so the count is defensible even if it feels dense.

The read-only query surface is well covered: current and historical incidents, event clustering, point/postcode/AGRN declaration lookups, gauge readings and summaries, calendar days, schema, and attribution. Minor gaps exist, such as no historical event-cluster listing and no direct declaration-by-ID endpoint, but agents can work around these.