AdLoop

The AI command center for Google Ads, GA4, and tracking code.

An MCP server that gives your AI assistant read + write access to Google Ads and GA4 — with safety guardrails that prevent accidental spend.

☁️ Skip the setup — use AdLoop Cloud (free plan, no card)  ·  or self-host: pip install adloop

[!TIP] AdLoop Cloud is the hosted version of this project, with a free plan that needs no credit card. Connect Google in two clicks and use the full toolset from claude.ai, ChatGPT, Claude Code, Cursor, or Gemini. No Google Cloud project, no developer token, no OAuth verification wait. EU-hosted, GDPR-first, DPA included.

📚 Documentation: docs.getadloop.com — setup guides per AI client, toolsets, the safety model, and troubleshooting for both editions.

---

Cloud or Self-Hosted?

Both versions run the same tools with the same safety model. The difference is who handles the plumbing:

	☁️ AdLoop Cloud	🛠️ Self-hosted (this repo)
Setup	Connect Google in two clicks	~5 min: own Google Cloud project + adloop init
Google Cloud project	Not needed	Required (free)
Ads developer token	Not needed	Required (from your MCC)
Works with	claude.ai, ChatGPT, Claude Code, Cursor, Gemini	Claude Code, Cursor, Claude Desktop, any local MCP client
Where your data flows	EU servers (Germany), GDPR-first, DPA included	100% your machine — nothing leaves it
Updates	Automatic	pip install -U adloop
Price	Free plan, no card; paid plans for more accounts and volume	Free forever (MIT)

Not sure? Start with Cloud — it's the fastest way to see what AdLoop can do, and it's the only way to use AdLoop from claude.ai or ChatGPT. Self-host when you want everything on your own machine or need to modify the code. And if you're here to hack on AdLoop itself: welcome, keep scrolling.

What It Solves

AdLoop exists because managing Google Ads alongside your code is a mess. These are the specific problems it handles:

• "My conversions dropped and I don't know why." AdLoop cross-references Ads clicks, GA4 sessions, and conversion events in one query. It detects whether the gap is from GDPR consent rejection, broken tracking, or an actual landing page problem — before you waste hours checking each dashboard separately.

• "I'm wasting ad spend on irrelevant searches." Pull your search terms report, identify the junk, and add negative keywords — all from a single conversation in your IDE. No context-switching to the Ads UI.

• "Is my tracking even working?" Compare the event names in your actual codebase against what GA4 is receiving. Find the mismatches: events you fire that GA4 never sees, events GA4 records that you didn't know about.

• "I need to create ads but the Google Ads UI is hostile." Draft responsive search ads, create campaigns, add keywords — all through natural language. Every change shows a preview first. Nothing goes live without your explicit confirmation. New ads and campaigns start paused.

• "My landing page gets paid traffic but nobody converts." AdLoop joins your ad final URLs with GA4 page-level data. See which pages get clicks but no conversions, which have high bounce rates, and which ones are orphaned from any ad campaign.

• "Are conversions even being tagged on every page?" AdLoop reads your live Google Tag Manager container, joins it against the events in your codebase and the events firing in GA4, and tells you exactly which conversions are being captured, which tags are paused, which page-scope filters are too narrow, and which codebase events have no tag at all — the kind of three-way audit GTM Preview can't give you in a single view.

• "I don't know if my EU consent setup is causing data gaps." In Europe, 30-70% of users reject analytics cookies. AdLoop accounts for this automatically — it won't diagnose a normal GDPR consent gap as broken tracking.

Built From Real Usage

Every tool exists because of an actual problem hit while running real Google Ads campaigns. The cross-reference tools exist because we kept manually asking the AI to "get Ads data, then get GA4 data, then compare them" — so we automated the join. The Broad Match + Manual CPC safety rule exists because the AI once created that exact combination and wasted budget. The GDPR consent awareness exists because the AI kept diagnosing normal EU cookie rejection as broken tracking.

The best features come from real workflows. If you're using AdLoop and find yourself wishing it could do something it can't, open an issue describing your situation — not just "add feature X" but "I was trying to do Y and couldn't because Z." The context matters more than the request.

All Tools

Quick start: pip install adloop or git clone https://github.com/kLOsk/adloop.git && cd adloop && uv sync && uv run adloop init — or zero setup on AdLoop Cloud

Diagnostics

Tool	What It Does
health_check	Test OAuth, GA4, and Ads connectivity in one call — actionable error messages if anything is broken. Also reports the pinned Google Ads API version and warns if a newer version is available.

GA4 Read Tools

Tool	What It Does
get_account_summaries	List GA4 accounts and properties
run_ga4_report	Custom reports — sessions, users, conversions, page performance
run_realtime_report	Live data — verify tracking fires after deploys
get_tracking_events	All configured events and their volume

Google Ads Read Tools

Tool	What It Does
list_accounts	Discover accessible Ads accounts
get_campaign_performance	Campaign metrics — impressions, clicks, cost, conversions, CPA
get_ad_performance	Ad copy analysis — headlines, descriptions, CTR
get_keyword_performance	Keywords — quality scores, competitive metrics
get_search_terms	What users actually searched before clicking
get_negative_keywords	List direct campaign-level negative keywords
get_negative_keyword_lists	List all shared negative keyword lists (SharedSets) — names, IDs, status, keyword count
get_negative_keyword_list_keywords	List the keywords inside a specific shared negative keyword list
get_negative_keyword_list_campaigns	List which campaigns a shared negative keyword list is attached to
get_recommendations	Google's auto-generated recommendations with type, estimated impact, and campaign context
get_pmax_performance	Performance Max campaign metrics with network breakdown + asset group ad strength
get_asset_performance	Per-asset details for PMax — field type, serving status, content
get_detailed_asset_performance	Top-performing asset combinations — which headline+description+image combos Google selects most
get_audience_performance	Audience segment performance — remarketing, in-market, affinity, demographics
get_demographic_targeting	List demographic criteria (age/gender/parental status/income) on an ad group or campaign
run_gaql	Arbitrary GAQL queries for anything else

Compact mode — get_campaign_performance, get_keyword_performance, get_search_terms, and get_ad_performance accept compact=true: account totals, breakdowns, top-10 rows, and pre-computed offender lists (zero-conversion spenders, low-QS keywords, negative-keyword candidates, thin RSAs) instead of every row. ~90% smaller responses — built for account audits so raw tables don't flood your AI's context.

Cross-Reference Tools (GA4 + Ads Combined)

These tools call both APIs internally and return unified results with auto-generated insights. They're the core of what makes AdLoop different from having separate GA4 and Ads tools.

Tool	What It Does
analyze_campaign_conversions	Maps Ads clicks → GA4 sessions → conversions per campaign. Detects GDPR consent gaps, computes real CPA, compares paid vs organic channels.
landing_page_analysis	Joins ad final URLs with GA4 page data. Shows conversion rate, bounce rate, and engagement per landing page. Flags pages with paid traffic but zero conversions.
attribution_check	Compares Ads-reported conversions vs GA4 events. Diagnoses whether discrepancies are from GDPR consent, attribution windows, or broken tracking.

Tracking Tools

Tool	What It Does
validate_tracking	Compare event names found in your codebase against what GA4 actually records. Returns matched, missing, and unexpected events with diagnostics.
generate_tracking_code	Generate ready-to-paste GA4 gtag JavaScript for any event, with recommended parameters for well-known events (sign_up, purchase, etc.) and optional trigger wrappers.

Google Tag Manager Tools

These tools read the live GTM container and join it with the codebase + GA4 to find tracking gaps that pure GA4 inspection can't catch — page-scoped triggers, paused tags, dynamic event names, brittle CSS selectors, and codebase events with no tag wired up at all.

Tool	What It Does
audit_event_coverage	The flagship. Three-way join: codebase events ↔ GTM tags ↔ GA4 actual fires. For each event name in expected_events, returns one of 10 statuses (ok, no_tag_no_fire, tag_paused, tag_active_but_not_firing, gtm_only_firing, ga4_only, etc.) plus auto-generated insights for the gaps.
list_gtm_accounts	Discover accessible GTM accounts
list_gtm_containers	List containers under an account — returns numeric container_id (needed by other tools), public GTM-XXXXXXX ID, and usage context (web/iOS/Android/server)
list_gtm_tags	Every tag in the live container with parsed event names and resolved firing/blocking trigger names
get_gtm_tag	Full raw config for a single tag — every parameter, firing/blocking triggers with filter conditions, priority, pause status, sampling
list_gtm_triggers	Every trigger with filter conditions parsed to readable text (e.g. {{Page Path}} contains service-promotions, {{Form ID}} NOT contains wf-form-...). Renders the negate flag explicitly.
get_gtm_trigger	Full trigger config + reverse lookup of every tag that uses it. Includes parsed element_visibility block (selector, on-screen ratio, firing frequency) for elementVisibility triggers and group_member_trigger_ids for triggerGroup types
list_gtm_variables	Custom variables (data layer, constants, JS) plus enabled built-in variables
list_gtm_workspaces	List drafts (workspaces) under a container — workspace IDs are needed by get_gtm_workspace_diff
get_gtm_workspace_diff	Drafted-but-not-published changes — common cause of "I edited a tag but nothing happened". Returns is_clean: true when nothing is pending.
list_gtm_versions	Publish history with version IDs and entity counts. Use to correlate a metric drop with a recent publish.
get_gtm_version	Full metadata + tag/trigger names for a single historical container version

Setup for GTM tools — Enable the Tag Manager API v2 in your GCP project, then add your AdLoop credentials' email (the OAuth user, or the service account email if using a service account) as a Read user on the GTM container under Admin → User Management. Service accounts pick up access on the next call. OAuth users upgrading from an earlier AdLoop version must re-authorize once: the GTM scope is new, so delete ~/.adloop/token.json and run any tool to re-consent — until then GTM tools return a permissions error.

Search Console Tools

Tool	What It Does
list_gsc_sites	List Search Console properties the connected account can access
run_gsc_report	Organic search analytics — clicks, impressions, CTR, position by query/page/country/device/date

Web Performance Tools

Tool	What It Does
analyze_page_speed	PageSpeed Insights for landing pages — Lighthouse score, Core Web Vitals, real-user CrUX data, top fixes. No OAuth needed (optional API key).

Merchant Center Tools

Tool	What It Does
list_merchant_accounts	Discover accessible Merchant Center accounts
get_merchant_feed_health	Feed health — approved/pending/disapproved counts per reporting context, top product issues with docs, account-level issues. Disapprovals silently starve Shopping/PMax.

Setup for Merchant Center tools — Enable the Merchant API in your GCP project (the Content API for Shopping is deprecated). The Merchant API has no read-only scope; AdLoop uses it strictly read-only. Upgrading OAuth users re-authorize once.

Setup for GSC tools — Enable the Search Console API in your GCP project. Upgrading OAuth users must re-authorize once for the new scope (delete ~/.adloop/token.json, run any tool). The killer combo: cross-reference organic queries with get_keyword_performance to find paid/organic cannibalization and untapped keyword opportunities.

Planning Tools

Tool	What It Does
discover_keywords	Discover new keyword ideas from seed keywords and/or a URL — with optional per-month search history + seasonality insights (include_monthly_volumes) using Google Ads Keyword Planner. Returns avg monthly searches, competition level, and top-of-page bid range.
estimate_budget	Forecast clicks, cost, and conversions for a set of keywords using Google Ads Keyword Planner. Supports geo/language targeting. Essential for budget planning before launching campaigns.

Google Ads Write Tools

All write operations follow a draft → preview → confirm workflow. Nothing executes without explicit approval.

Tool	What It Does
draft_campaign	Create a full campaign structure — budget + campaign (PAUSED) + ad group + optional keywords. Supports Search partners, display expansion, and max_cpc for either MANUAL_CPC initial ad-group bids or TARGET_SPEND (Maximize Clicks) CPC caps.
update_campaign	Modify existing campaign settings — bidding, budget, geo/language targeting, Search partners, display expansion, and TARGET_SPEND (Maximize Clicks) max_cpc caps.
draft_ad_group	Create a paused SEARCH_STANDARD ad group inside an existing campaign, with optional MANUAL_CPC max_cpc.
update_ad_group	Update an ad group name and/or MANUAL_CPC max_cpc. Use pause_entity / enable_entity for ad-group status changes.
draft_responsive_search_ad	Create RSA preview (3-15 headlines ≤30 chars, 2-4 descriptions ≤90 chars). Warns if headline/description count is below best practice.
draft_callouts	Create campaign callout assets from 1-25 character text snippets.
draft_structured_snippets	Create campaign structured snippet assets using official header values and 3-10 snippet values.
draft_image_assets	Create campaign image assets from local PNG, JPEG, or GIF files.
draft_keywords	Propose keyword additions with match types. Proactively checks bidding strategy — blocks BROAD match on Manual CPC campaigns.
add_negative_keywords	Propose negative keywords directly on a campaign
add_negative_locations	Propose negative geo exclusions on a campaign — exclude cities/regions while keeping broader positive targets
draft_key_event	Mark a GA4 event as a key event (conversion) — the fix for "fires but isn't tracked as a conversion"
draft_demographic_targeting	Propose demographic criteria (age, gender, parental status, income) — exclusions by default
propose_negative_keyword_list	Draft a shared negative keyword list (SharedSet) and attach it to a campaign — reusable across multiple campaigns
pause_entity	Pause a campaign, ad group, ad, or keyword
enable_entity	Re-enable a paused entity
remove_entity	Permanently remove an entity (irreversible — prefers pause). Supports keywords, negative keywords, ads, ad groups, campaigns.
confirm_and_apply	Execute a previously previewed change

Orchestration Rules

AdLoop ships with orchestration rules that teach the AI how to combine these tools — marketing workflows, GAQL syntax, safety protocols, GDPR awareness, and best practices. Without rules, the AI has tools but doesn't know the playbook.

• Cursor: .cursor/rules/adloop.mdc (canonical source)
• Claude Code: .claude/rules/adloop.md (synced from Cursor rules via scripts/sync-rules.py)

The rules include:

• Orchestration patterns for common workflows (performance review, conversion diagnosis, campaign creation, negative keyword hygiene, keyword discovery, tracking validation, budget planning, landing page analysis)
• GAQL quick reference with syntax, common queries, and gotchas
• Safety rules including Broad Match + Manual CPC prevention and pre-write validation
• Ad copy character limit guidance (30-char headlines are shorter than you think)
• GDPR consent awareness to prevent false tracking diagnoses in EU markets

Slash Commands (Claude Code)

AdLoop includes pre-built slash commands in .claude/commands/ for common workflows:

Command	What It Does
/analyze-performance	Full performance review across Google Ads + GA4
/create-ad	Create a responsive search ad with safety checks
/diagnose-tracking	Diagnose tracking and conversion issues
/optimize-campaign	Full optimization checklist for a campaign
/create-campaign	Create a new search campaign with budget estimation
/budget-plan	Estimate budget for keywords via Keyword Planner

Safety Model

AdLoop manages real ad spend, so safety is not optional.

• Two-step writes. Every mutation returns a preview first. A separate confirm_and_apply call is required to execute.
• Dry-run by default. Even confirm_and_apply defaults to dry_run=true. Real changes require explicit dry_run=false.
• Two-phase apply (optional). With safety.two_phase_apply: true, confirm_and_apply refuses dry_run=false until the plan has completed one dry-run pass — preview-then-apply becomes server-enforced instead of a convention.
• Budget caps. Configurable maximum daily budget — the server rejects anything above the cap.
• Audit log. Every operation (including dry runs) is logged to ~/.adloop/audit.log.
• New campaigns and ads are PAUSED. Nothing goes live without manual enablement.
• Destructive ops require double confirmation. Removing entities or large budget increases trigger extra warnings.
• Broad Match + Manual CPC blocked. The #1 cause of wasted ad spend is automatically prevented — draft_keywords refuses to add BROAD match keywords to campaigns without Smart Bidding.
• Pre-write validation. Before any write, the AI checks bidding strategy, conversion tracking status, and quality scores. If the campaign is fundamentally broken, AdLoop warns you instead of making things worse.
• Structured error handling. All tools return actionable error messages with hints instead of raw exceptions. Auth errors include specific re-authorization steps.
• API version pinning. The Google Ads API version is pinned to prevent silent breaking changes from library updates. health_check warns when a newer version is available.
• Ask mode compatibility. Read tools declare readOnlyHint so they work in Cursor's Ask mode without switching to Agent mode.

Setup

AdLoop uses your own (free) Google Cloud project for OAuth. The adloop init wizard walks you through it — a one-time setup of about 5 minutes, with no shared user caps and no waiting on anyone's verification review. AdLoop does not ship built-in OAuth credentials.

Prefer zero setup? AdLoop Cloud is the hosted version: connect Google in two clicks — no Cloud project, no developer token, EU-hosted.

(Upgrading from ≤0.9 with built-in credentials? Those sign-ins were retired in 0.10 — run adloop init once to switch to your own project.)

[!IMPORTANT] Seeing deleted_client: The OAuth client was deleted. or invalid_client? The shared Google Cloud project behind AdLoop ≤0.9's bundled credentials has been shut down, so its stored sign-ins no longer refresh. Two ways forward: AdLoop Cloud (connect Google in two clicks, nothing to configure) or stay self-hosted with pip install -U adloop && adloop init to set up your own free Google Cloud project. Details in the pinned issue (#49).

Install

From PyPI:

pip install adloop
adloop init

From source:

git clone https://github.com/kLOsk/adloop.git
cd adloop
uv sync
uv run adloop init

What adloop init does

The wizard walks you through:

1. Google Cloud setup — creates a project, enables the three APIs, generates an OAuth client (see Custom Google Cloud Project Setup below for the exact steps the wizard refers you to)
2. Developer token — from your Google Ads MCC (API Center)
3. MCC Account ID — your Manager Account ID (top bar in the MCC UI)
4. OAuth sign-in — opens a browser to sign in with Google (or prints a URL for headless servers)
5. Auto-discovers your accounts — finds your GA4 properties and Ads accounts automatically
6. Optional services — pin a GTM container, a Search Console property (both auto-discovered too), and a PageSpeed API key; skip any of them with Enter
7. Safety defaults — budget cap and dry-run preference
8. Toolsets — optionally expose only part of the tool catalog to your AI client (see Toolsets)
9. Editor config snippets — prints MCP configuration for both Cursor and Claude Code, including your toolset selection

Requirements

• Python 3.11+
• A Google Ads account with an MCC (Manager Account)
• A Google Ads Developer Token (see below)

Google Ads Developer Token

A developer token is always required. Your OAuth client handles Google sign-in; the developer token is a separate key that grants API access to your Google Ads data.

1. Create an MCC (free) at ads.google.com/home/tools/manager-accounts if you don't have one. Link your regular Google Ads account to it.
2. In the MCC, go to Tools & Settings → API Center
3. Your developer token is shown there. Copy it — the wizard will ask for it.

Access levels — your token's access level determines what it can do: