{"id":"klosk-adloop","name":"adloop","af_score":66.5,"security_score":67.2,"reliability_score":42.5,"what_it_does":"adloop is a Python MCP server that lets an AI assistant read and (safely) write to Google Ads and GA4. It provides numerous MCP tools for diagnostics, GA4/Ads reporting, cross-referencing Ads vs GA4 (including attribution and landing-page analysis), tracking validation, and a draft→preview→confirm workflow for creating/updating Ads assets with guardrails (budget caps, pausing new entities, and double-confirmation for destructive actions).","best_when":"Teams want an AI-driven workflow inside Cursor/Claude Code to compare Ads vs GA4 and to propose Ads changes with explicit, confirm-and-apply guardrails.","avoid_when":"You cannot or will not provide OAuth credentials and developer token access, or you require highly formalized change management beyond a dry_run/confirm workflow.","last_evaluated":"2026-03-30T15:23:29.260123+00:00","has_mcp":true,"has_api":false,"auth_methods":["OAuth (desktop app flow) using Google OAuth client credentials","Service account key JSON (supported per README)"],"has_free_tier":false,"known_gotchas":["Write operations require explicit confirm_and_apply with dry_run=false for real changes; agents must avoid assuming writes happen automatically after draft/preview.","Budget caps and pausing defaults may cause confusion if the agent expects immediate live spending/serving.","Service-account vs OAuth credential types may differ operationally; ensure the correct credential path and account permissions in advance."],"error_quality":null}