Related Servers

Most tools have a clearly distinct purpose and the descriptions actively cross-reference siblings (e.g. propose_negative_keyword_list vs add_negative_keywords vs add_to_negative_keyword_list), which sharply limits misselection. However, several name-level clusters remain confusable at a glance (get_negative_keywords vs get_negative_keyword_lists vs get_negative_keyword_list_keywords; get_brand_lists vs get_brand_list_brands vs get_brand_list_campaigns; analyze_campaign_conversions vs attribution_check vs audit_event_coverage), and draft_ad_group/draft_campaign optionally absorb keywords that draft_keywords also adds.

Names are uniformly snake_case and follow a verb_noun shape, so the surface is readable and predictable. The verb choice is heterogeneous, though: the same 'preview then confirm_and_apply' pattern is exposed as draft_*, add_*, propose_*, attach_*, detach_* and update_*, and there are outliers like health_check and confirm_and_apply, which is a minor deviation rather than a break in convention.

137 tools is far more than one server should expose, well past the 50+ threshold in the rubric. The scope genuinely spans Google Ads, Reddit Ads, GA4, GTM, GSC, Merchant Center and PageSpeed, which justifies breadth, but the read/write/discovery surface is fragmented into so many near-adjacent tools that the set is heavy and hard to navigate.

Coverage is extensive across every advertised platform: discovery, full read/report coverage, and a consistent draft → confirm_and_apply write pipeline spanning CRUD for campaigns, ad groups, ads, keywords, audiences, brand lists, negatives, conversions and GTM entities, plus diagnostics and uploads. The few gaps (e.g. no campaign creation for Display/Video/PMax, no GA4 property administration) are deliberate refusals rather than missing operations.