Scored across 48 tools
Each tool targets a distinct resource+action, and the set cleanly separates monitored dimensions (get_tech_trust_dashboard) from live one-off scans (start_trust_signals_scan, check_sitemap) and from briefing/ticketing. The main friction is the parallel naming of near-identical access patterns (get_content_run_detail vs get_ai_visibility_check_detail) and the cluster of three similar-sounding scans, which descriptions laboriously disambiguate but a rushed agent could still mix up.
Consistent snake_case verb_noun throughout: list_*, get_*, start_*, create_*, update_*, move_*, delete_*, add_*, check_*, fetch_*. The dimension families follow a perfectly predictable template (<dimension>_dashboard / _history / _run_detail), and even the runId-vs-checkId distinction is reflected deliberately in the '_check_detail' suffix.
48 tools is heavy and well past the comfortable range, even for a platform with six monitoring dimensions, a briefing subsystem, and a full ticketing board. Almost every tool has a real, non-duplicative role, so nothing is obviously padding, but the surface is large enough that discovery and selection cost is significant.
Coverage is broad: per-dimension dashboards, paginated histories and run details, live scans, alerts, briefings, and full ticket lifecycle including comments, labels and assignees. Gaps are minor and partly deliberate — schedules can be listed but not modified, labels cannot be created, and competitors/projects are read-only — leaving a few dead ends an agent must work around.