Adako: Google Ads, Meta Ads & Linkedin Ads MCP

The router/write-router split (google_ads vs google_ads_write, etc.) is deliberate and each description explicitly says what it does not do, which keeps most tools apart. However, there is real overlap in the discovery/account layer: get_connections_status vs list_connected_accounts, list_pending_proposals vs monitoring's list_pending_actions (two different 'pending' concepts), and list_what_i_can_do vs start_here vs search_tools all answer 'what can I do here?'.

Predominantly consistent snake_case verb_noun, with predictable platform prefixes (google_*, meta_*, linkedin_*, tiktok_*, chatgpt_*). Minor deviations exist: generic verbs differ across platforms (meta_pause_entity vs linkedin_pause_campaign), and top-level system tools mix get_/list_/verb forms (get_connections_status vs list_connected_accounts), but nothing is chaotic.

29 top-level tools is already on the heavy end, and each platform router fans out to many more (13 ChatGPT, 38 Google, 17 LinkedIn, 16 Meta, 12 TikTok, plus monitoring/diagnostics sub-tools, over 100 total). The router-with-free-list_tools pattern is a reasonable mitigation, but the surface is large enough that an agent must constantly do lookup calls to navigate it.

Coverage is broad and mature: reads, writes, approval/proposal lifecycle, diagnostics, monitoring, reporting and connection management across five ad platforms. The main gap is intentional and documented (no campaign/ad-group/ad deletion), and some lifecycle operations are asymmetric (delete only for creatives on LinkedIn/Meta), but agents can work around these.