Adrails
Server Details
Run Meta Ads and Google Ads for your Shopify store from Claude, ChatGPT or Cursor, with the store's own Shopify, Stripe or Klaviyo revenue next to ad spend. New campaigns are created paused.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- OAuth
- Works in Glama
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- Repository
- devthomy/adrails-mcp
- GitHub Stars
- 0
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 47 tools
The read/analyse/review clusters overlap heavily: performance_period, reporting_overview, metrics_daily, performance_diagnose, business_review, planning_calculate and metric_compare all answer numeric performance questions, and changes_review, actions_recent, findings_review and portfolio_review all review recorded changes. The unusually long descriptions do draw careful boundaries (which level, which window, which method), but an agent must read a lot of prose to choose correctly, and several pairs (creative_insights vs creative_review vs creative_market_evidence, knowledge_propose vs knowledge_skills) still require judgement.
Almost everything is lowercase snake_case in a noun_action shape (accounts_list, ads_export, budget_recommend, store_revenue, workflows_list), which is predictable. The deviations are minor: singular/plural inconsistency on the campaign family (campaign_drafts, campaign_media, campaign_templates vs campaigns_setup, campaigns_structure) and noun-only names (site_brief, jobs_status, launch_brief).
47 tools is far above a comfortable surface and the categories sprawl beyond ad operations into commerce, SaaS metrics, Google Drive, workflows, knowledge and calculation utilities. While the underlying domain is genuinely broad, the set is heavy enough that tool selection becomes a search problem rather than a choice among obvious options.
Read coverage is extensive (accounts, structure, performance, creatives, leads, revenue, targeting, libraries), but the surface references tools that are absent from the list: meta_edit, google_ads_edit, google_ads_build, meta_audience_create, lead_form_create, campaigns_launch and campaigns_publish_draft are all cited in descriptions yet missing, leaving dead ends wherever a read tool says 'these ids go to X'. Draft creation and validation exist, but publishing and editing must happen elsewhere, so the lifecycle is only partly closed.
Available Tools
47 toolsaccounts_listAccounts listBRead-onlyIdempotentInspect
Lists connected ad accounts, timezone, currency, freshness and enabled state.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
actions_recentActions recentARead-onlyIdempotentInspect
Recorded actions with ids, evidence, execution state, undo availability and 24 hour, 3 day and 7 day observations: the source for undoing or reporting changes. since and accountIds scope a day or a period and add each platform's own log, Meta's activity log or Google Ads' change history (the last 30 days Google keeps, with who changed what in Google Ads, Editor, scripts, rules or Google's auto-applied recommendations), as one report over every account; by keeps one person's changes: me (whoever connected the platform), others (everyone else) or a name, an email on Google. When the log cannot tell who the person is, the result names the people it holds, for a new call with by set to one of them. A change report's table carries its CSV and PDF as links signed for the connected person, valid for one hour. Without showTable it returns rows for citing relevant applied changes with their timing, without causal claims. Paginated with nextOffset over the same scope.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
ads_exportAds exportARead-onlyIdempotentInspect
Builds a file of each account's ads over a date range, Meta's and Google Ads', as CSV and PDF: the Ads Manager columns, each Meta ad's share of spend by age, gender, placement and country, and the ad text. filters.by also takes region, dma, frequency, product or user_segment, each alone: region, DMA and product rows hold no purchases, and frequency rows hold reach alone. A Google account gives its ads and the asset groups Performance Max has instead, or with filters.level keyword, asset, product, search_term, landing_page, placement or audience its keywords, its assets (Performance Max's too, with their status), its Shopping and Performance Max products with their attributes, its search terms (Performance Max's too), the pages people landed on, where its ads ran or the audiences its campaigns and ad groups add; filters.by a custom label (custom_label_0), category, product type or brand gives one product row per value, and filters.label0 to label4, category, type or brand keep those values. Rows carrying Google's cart data add Google orders, revenue, cost of goods, gross profit, margin and return on profit. The result holds the file's title, the accounts in it, up to 5 preview rows and its file links, signed for the connected person and valid for one hour; the file is read from the ad platform when a link is opened. A file holds each account's top 15 ads by spend, top sets up to 50, and every ad that spent is included when filters.all is true or the userStatement asks for all of them ("all my ads", "every ad"); a period and the columns named in the userStatement shape the file as they do in the app. kind "comments" (or a userStatement asking for the comments) builds instead a file of every comment and reply under the Facebook and Instagram posts of the Meta ads that spent over the dates, the last 30 days when none are given, narrowed by filters.campaignIds or filters.adIds (Meta ids) or by campaign and ad names: CSV with one row per comment (ad account, campaign, ad set, ad, surface, post id and link, comment id, parent comment id, author as Meta returns it, text, ISO time, likes, replies, hidden), PDF, and a text for Claude grouped by ad with its spend, up to 20,000 comments. Where the Meta connection lacks a comment permission, the result and the file name the missing permission and the other surface is still read.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
audience_estimateAudience estimateBRead-onlyIdempotentInspect
Estimates Meta reach for proposed targeting using the existing account audience service. An estimate is not measured performance.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
brand_read_siteBrand read siteARead-onlyIdempotentInspect
Reads one of the workspace's stores: a connected Shopify store through Shopify (store details, best-selling products) and its public pages (about, FAQ, shipping, returns, contact), else the url on the public web. With several connected stores, pass store (its name or domain); without it the call lists them. Returns each source's address and text, and the store's logo, colors and fonts from its theme, to write a brand profile from (overview, story, products, audience, positioning, voice and tone, promises, objections, claims to avoid), each line resting on a source. Runs no model and saves nothing; the returned link is where the brand profile is reviewed.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
budget_recommendBudget recommendARead-onlyIdempotentInspect
Evaluates the budget-owning entities of one selected account as maintain, increase, reduce or investigate (the scaling-decision method). Without entityId and entityType it finds and reviews the actual budget owners, paginated; with both it reviews one campaign or ad set. The event and any cost cap or minimum ROAS target come from the actual ad set settings, not from generic results or names. Compares the latest two complete seven-day windows in the account time zone and reports gaps, evidence and a review point. A missing target stays missing, and delivery targets are not profit. Recent changes and business constraints are not part of this read: actions_recent and knowledge_skills hold them. Proposes no amount and writes nothing. responseMode=answer returns a report; evidence returns the underlying data.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
business_reviewBusiness reviewARead-onlyIdempotentInspect
Explains or calculates unit economics, revenue reconciliation or budget pacing from explicit inputs. Missing inputs are optional and stay missing: no example, cost, target or date is added. Every supplied number carries a verbatim quote from a user message or a source tool result; assistant examples and Knowledge instructions are not accepted as sources, and the tool checks the quotes, not their business meaning. A generic AOV is not treated as verified net first-order revenue. Pacing counts complete elapsed days only. Figures are combined only when their currency, definition and window match. responseMode=answer returns a report; evidence returns the underlying data. No writes or forecasts.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
campaign_draftsCampaign draftsADestructiveInspect
Manages unpublished workspace campaign drafts. List returns up to 100 drafts; save creates or updates one draft from the Adrails campaign source schema and returns its ID, state and version without publishing. validate checks a saved draft; campaigns_publish_draft publishes a validated draft after its own approval. Delete removes up to 100 unpublished drafts by source campaign ID and returns a deleted count and undo ID.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
campaign_mediaCampaign mediaBInspect
Lists an ad account's platform media. Upload sends nothing from the AI client: it returns the Adrails link where the person uploads the file in Campaigns. Upload does not launch ads.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
campaigns_setupCampaigns setupARead-onlyIdempotentInspect
Reads the Meta account settings a new launch or a read-only launch check needs: the selected Meta accounts with internal accountId, currency, snapshot freshness and available Page IDs, the account's datasets (pixels) as the editor lists them, the datasets its ad sets optimize for (adSetDatasetIds), and image hashes. Edits to what already runs (budget, bid, status, placements, audience, dates, copy) take their ids from campaigns_structure. Candidates listed here are not confirmed selections, permission or event checks, or a passed preflight. Without an explicitly scoped account it returns the accounts to pick from. An empty asset list means media has to be uploaded, or an identity connected, in Adrails Campaigns. Read only.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
campaigns_structureCampaigns structureARead-onlyIdempotentInspect
Reads, from the last sync, how an ad account's campaigns are built on any platform: every campaign with its type, status, budget and bidding; every ad group or ad set with its keywords (ids, match types, status, quality scores, review, and spend and conversions over the sync's days, the ones that spent most kept when a campaign is large) and, for Performance Max asset groups, each text with its asset id and the assets Google disapproved; for Shopping ad groups and Performance Max asset groups, the products their listing groups include and exclude, by brand, category, product type, custom label or item id; on Google, each campaign's own negative keywords, the shared negative lists it uses and the extensions Google holds back, and its controls: the portfolio strategy it bids with and the target actually in force with what sets it, Google's suggested budget, the placement exclusion and brand lists it uses, labels, ad schedule and device bid adjustments; at account level the portfolio strategies, every shared list with its first entries, what the account excludes and its user lists with their sizes; every ad with its format, headlines, descriptions, paths, final URL, pictures and review. On Meta, every ad set's tracking (destinationType, the conversion location: WEBSITE or ON_AD for an instant form; optimizationGoal; the datasetId and event it optimizes for) and every ad's urlParameters (null when it has none), finalUrl, leadFormId and the datasets its tracking reports to. Its ids are the ones meta_edit, google_ads_edit and google_ads_build take for existing entities. campaignId narrows it to one campaign.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
campaign_templatesCampaign templatesADestructiveInspect
Manages reusable campaign, ad set or ad templates in the current workspace. List returns template IDs, names, channels, types, payloads and update timestamps, with optional channel and type filters. Create requires a fixed type; update replaces its name, channel and payload while retaining that type. Delete returns an undo ID and leaves existing campaign drafts and published ads unchanged.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
changes_reviewChanges reviewARead-onlyIdempotentInspect
Reviews recorded actions with their actual application dates, baseline and observation eligibility (the change-review method). Pending, failed and skipped actions are kept apart from measured results, and rolling snapshots are not presented as causal lift. Scoped to one selected account, optionally one action; paginated. responseMode=answer returns a report; evidence returns the data for broader tasks or several pages or accounts. No writes and no saved learning.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
channels_manageChannels manageADestructiveInspect
Lists or manages automation delivery channels. From this connection a channel can only be created or pointed at the email addresses of verified members of the workspace; Slack, Discord, Telegram and any other recipient are added by a person in Adrails. Renaming, testing and deleting work for every channel. Email channels save recipients without sending a test email. Channel targets are private.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
creative_insightsCreative insightsBRead-onlyIdempotentInspect
Groups ads by their actual image or video asset, across ads and accounts. Returns winners, fatigue, repeated usage, performance, video hook and hold rates, and missing angle, offer, product, persona or creator metadata, as the basis for creative variants and hook comparisons.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
creative_market_evidenceCreative market evidenceARead-onlyIdempotentInspect
Per-ad performance, copy and media ids, the parent ad set's target countries and the snapshot period for 1 to 3 adIds of any age, with video retention, plays, hook, hold and placements, and, for an ad with several primary texts, headlines, images or videos, each one's spend, CTR, purchases, CPA and ROAS from Meta's asset breakdown (approximate per asset, not adding up to the ad's total): the evidence for reuse, localization, new-market tests or which asset wins inside an ad. It is not a country delivery breakdown: results across several countries are not split by country, and visuals are not inspected.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
creative_reviewCreative reviewARead-onlyIdempotentInspect
Reviews the evidence of selected ads for the creative-map, creative-fatigue, creative-roadmap and market-test methods: available copy, asset IDs, configured goal, period and limits. Fatigue, winning status, product promises and sample thresholds appear only where the data shows them. focus=copy_test proposes one text-order test that reuses observed wording and keeps every other variable. focus=market_transfer keeps the source asset and copy and uses only the destination the person named. A source ad is a control candidate, not a validated winner. responseMode=answer returns a report; evidence returns the underlying data. Produces no asset and prepares no campaign.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
drive_folderDrive folderARead-onlyIdempotentInspect
Reads the creatives in a Google Drive folder shared with the workspace's Google address. With folderId (the folder_id of a drive_folder mention, or one a search returned) it lists every image and video down to three levels, each with its path, the ratio Drive measured (9:16, 4:5, 1:1, 1.91:1), pixel size, video duration and file size, then Adrails' default grouping into ads: each file at the top of the folder is one ad, and each subfolder is one ad whose files are the same creative in several ratios. Without folderId it returns the folders whose name contains query. The listing keeps the durations, ratios and names that show when files are organized another way, such as ratio names in file names or raw and logo files. An empty result means no folder shared with the workspace's Google address matches. Read only.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
editor_settings_referenceEditor settings referenceARead-onlyIdempotentInspect
Returns every setting of the Meta campaign editor that launch_brief's metaSettings accepts, with its allowed values and format; tool=meta_edit, meta_audience_create or lead_form_create returns that tool's format instead. Read only.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
findings_reviewFindings reviewARead-onlyIdempotentInspect
Reviews what the rule engine found on the selected Meta, Google Ads and TikTok campaigns against the workspace's own standard. Each finding carries observed evidence, its snapshot date, the campaign objective, the ad set or ad group event and destination, currency, budget owner and a classification: confirmed issue, policy deviation, hypothesis, platform context, not applicable or unknown. Severity is not causality. A purchase preference does not override traffic or lead goals; automatic placements and missing URL tags are not evidence of wasted spend or lost attribution, and CPM or CTR signals alone are not evidence of fatigue. Campaigns without findings are covered by reporting_overview.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
google_ads_libraryGoogle ads libraryARead-onlyIdempotentInspect
Reads a Google Ads account's pictures (assetId, address and size, the newest that fit with the account's count, or those whose name holds the words of search), its YouTube videos, its audience lists and its audiences. Pictures and videos are reused by id in google_ads_build; audience lists are what an ad group's audienceLists name, and audiences name a signal for a Performance Max asset group. google_ads_build accepts only pictures listed here or given by the person.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
google_keyword_plannerGoogle keyword plannerARead-onlyIdempotentInspect
Reads Google's Keyword Planner for a Google Ads account. mode volumes returns how often each keyword is searched a month and its top of page bid range; mode ideas returns keywords suggested by a few words or a page (https url), most searched first, with the same figures. Counts cover the countries and language given, or everywhere. Keywords Google gives no figure for are listed apart. Gives a search campaign keywords measured on real searches. Read only.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
integrations_catalogIntegrations catalogBRead-onlyIdempotentInspect
Lists implemented providers, available data and connection requirements.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
integrations_detailIntegrations detailBRead-onlyIdempotentInspect
Reads one connection's imported data coverage, sync streams and trend.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
integrations_statusIntegrations statusARead-onlyIdempotentInspect
What the workspace has connected, what each source can do, and how fresh each data stream is, including failing syncs and sources not connected yet.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
jobs_statusJobs statusARead-onlyIdempotentInspect
Reads persisted background job, publication, action or workflow state. Pending, applied and failed are distinct states; a pending job is not complete.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
knowledge_proposeKnowledge proposeARead-onlyIdempotentInspect
Checks a durable instruction or a reusable request (shortcut) the person stated, against what Knowledge already holds. Saves nothing: returns the shortcut to save with knowledge_manage once the person says yes, or the link where they add an instruction.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
knowledge_skillsKnowledge skillsARead-onlyIdempotentInspect
Reads the workspace's methods (skills) and documents for specialist work. command with a known route key (budget-pacing, unit-economics, revenue-reconciliation, market-test and others) returns a method's full body in one call; query searches, and a returned skillId or documentId reads one entry. Strategic proposals, read-only plans and requests not to launch each have a method. Lists return metadata with nextOffset. Workspace scope and approvals apply.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
launch_briefLaunch briefAInspect
Prepares or continues a Meta launch brief. userStatement carries the user's request or answer for this launch in their own words, and input carries only confirmed choices. Returns the open questions with their labels and named options, and maps names only to IDs it supplied; each call with new answers updates the brief. The account, budget and launch mode come from the person, not from defaults. A Leads launch on an instant form takes the form's id from userStatement or from metaSettings.ad.leadFormId; targeting_options kind=lead-form lists a Page's forms. A completed brief goes to campaigns_launch, which returns an Adrails approval link: nothing goes live or spends without approval in the browser.
Inputs: input (required), userStatement (required), accountIds, idempotencyKey.
lead_sheet_sync_statusLead sheet sync statusARead-onlyIdempotentInspect
Lists the Google Sheets leads are sent to in the workspace's accounts (syncId, the campaign, ad set or ad, the sheet's link, leads sent, the last error) and the connected person's Google sign-ins, with the sign-in link when none can create a sheet. A sheet opens for whoever its owner shared it with in Google.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
leads_readLeads readARead-onlyIdempotentInspect
Counts the leads of a Meta campaign, ad set, ad or instant form (scope and its Meta id) from since to until, in the ad account's days: the total, by day, by form, by ad set and by ad, with the address of the Leads tab. The leads themselves (when each came, every answer, the form and the ad) are a CSV at file, signed for the connected person, valid for one hour and opened once signed in to Adrails. The result holds no name, email, phone or answer: Meta is asked for each lead's id, time and ad alone. A connection without the leads permissions, or a Page the business connection did not grant, returns the Leads tab's sentence saying what grants it.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
metric_compareMetric compareARead-onlyIdempotentInspect
Compares one observed metric with a positive target and returns the signed and relative difference, the attainment percent, and whether the metric is above, below or equal to the target. Meant for requests for a numeric difference, a percentage gap or an attainment percentage; it does not explain causes or trends.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
metrics_dailyMetrics dailyARead-onlyIdempotentInspect
Daily series measured by the connected integrations: traffic, sessions, conversion rate, sales and average order value from the store; emails received, opened and clicked from marketing tools; net profit, CPA and ROAS from analytics tools. Without metrics it returns everything measured over the period, which lists the available metric names. breakdown splits by product, traffic source or country instead of totals. Ad campaign figures are not included; reporting_overview returns them.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
performance_diagnosePerformance diagnoseARead-onlyIdempotentInspect
Compares two equal-length, non-overlapping historical periods from one saved account snapshot (the performance-diagnosis method). Takes the exact account ID and dates of the request, computes coverage and changes, and separates facts, unknowns and the next check. Missing rows are not counted as zero, and generic results are not counted as purchases. responseMode=answer returns a verified report; evidence returns the underlying data. No writes.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
performance_periodPerformance periodARead-onlyIdempotentInspect
Reads one ad account's provider figures for an exact date range on Meta, Google Ads or TikTok, at account, campaign, ad set (ad group or asset group on Google) or ad level: spend, results, revenue, ROAS, CPA, impressions, clicks, CTR, CPC, CPM, and reach and frequency on Meta. compareSince and compareUntil read a second range and give each entity's change. Covers any period other than the synced window, trends, and before-and-after reads. Dates are YYYY-MM-DD in the account time zone, 400 days at most. campaignIds narrows the read; entityIds keeps only those rows; breakdown cuts one period by age, gender, country, placement (the network on Google), device or hour, and on Google Ads by region or city too (people located there apart from people interested in the place, with the Shopify store's sales by shipping region beside a region), and by parental status (parental) or household income (income), by day (with each day's impression shares), day_of_week, click_type, slot (top of the page or other) or conversion_lag (days from the ad to the conversion). Google rows add the impression share family (lost impression share to budget or rank, top and absolute top, click share, Display), all, view-through, cross-device and by-conversion-date conversions, invalid clicks, and a video's TrueView views and quartiles. On Google Ads it also reads level keyword (match type, status, quality score and its three parts, impression shares), search_term (Search's and Performance Max's terms, the keywords that matched, where Google's match came from: the advertiser's keyword, AI Max or Performance Max, and whether a negative already excludes the term), asset (each asset of the ads, asset groups, campaigns and account with its status, why it is limited, whether Google made it and its figures, and what Google asks to add to an asset group for its ad strength), landing_page (each page people landed on, the pages final URL expansion chose, Google's mobile and speed scores, and the Shopify product a page shows with the orders Google's visits brought it), placement (each site, app and YouTube channel Display, YouTube, Demand Gen and Performance Max ran on, with the ones that look like junk said; nothing is excluded), product (Shopping and Performance Max products with their custom labels, category, product type, brand, channel and condition, Google's cart data where the account sends it: orders, revenue, cost of goods, gross profit and return on profit, the products bought in the same orders, the Shopify product behind each feed item and what the store's orders tied to Google bought of it, the products losing money after their spend by each source, the feed's refused products with their reasons and the Merchant Center link; tag groups the products by label0 to label4, category, type, brand, channel or condition) conversion_action (what each action recorded and its last day, its last conversion and tag fire ever, beside Adrails' own Data Manager uploads of store purchases) and audience (each remarketing or customer list, interest, custom or combined audience a campaign or an ad group adds, with its figures, whether it targets or only observes, its bid adjustment and a list's sizes on Search and Display). A Google campaign read adds each campaign's cart data and its new and returning customers, by Google's classification and by the store's orders tied to Google, side by side. A read gives 40 rows by spend, or 8 per group when it compares, says which rows of how many it gives, and offset reads the next ones; an entity without delivery has no row. metricSet=additive serves spend, conversion, revenue and efficiency questions and can reuse fresh campaign daily data; standard adds reach, frequency and fatigue metrics, whose unique counts need an exact period read. Meta figures count each ad set's own attribution setting, as Ads Manager does; a userStatement naming a window ("7-day click only", "1-day click", "without views", "1-day view", "engaged view") counts Meta's purchases and revenue in it, and one asking click versus view ("view-through") adds both windows beside Ads Manager's figure; the result names the window. When the call's userStatement asks what moved on the ads ("what moved this week", "which ads are dying"), it returns the report of what moved instead, at ad level from each Meta account's latest sync: the ads spending more, less, stopped, dying slowly or at a low ROAS, against the period before. On Meta, a userStatement asking by region or state, DMA, frequency, catalog product or customer segment (new against existing customers) cuts the period that way instead: region and DMA come without purchases, and region sits beside the store's net sales by shipping region, matched by name within the store's countries, when a store sells for the account; frequency gives reach per bucket; product names each catalog product, the product set's purchases and the store product tied to it; customer segments are those of Advantage+ sales campaigns, and the result says when the account defines no existing customers.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
planning_calculatePlanning calculateARead-onlyIdempotentInspect
Deterministic budget pacing, allocation totals or first-order acquisition unit economics, from verified tool inputs or hypothetical inputs the person supplied, each with its source and exact period in evidence, in one currency. Missing budgets, costs, time windows and target contributions stay missing. Returns the arithmetic and its limitations: not a forecast, verified profitability, an optimized allocation or a spending approval. Reads no account and changes nothing.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
portfolio_reviewPortfolio reviewARead-onlyIdempotentInspect
Advises across the selected accounts and campaigns after an audit, from each campaign's actual objective, ad set events, budget ownership, current-period coverage and freshly evaluated rules. Separates provider issues, policy differences, hypotheses and unknowns, includes campaigns with no findings, keeps intentional traffic, lead, purchase and intermediate goals, and synthesizes at most three supported priorities. CPA, causal defects, pause durations, QA purposes and budget or test plans appear only where the data supports them. Paginated; no writes. responseMode=answer returns a report; evidence returns the data for an additional deliverable such as a creative brief or an executable plan.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
reporting_overviewReporting overviewARead-onlyIdempotentInspect
Spend, attributed revenue and named conversion events for the workspace's connected ad accounts on Meta, Google Ads and TikTok. Without dates it returns each account's latest synchronized window with entity settings and status; with entityIds, those exact campaigns, ad groups or ads with their platform and ids, and for a campaign its last seven days against the seven before. lastDays=N covers the last N complete calendar days (today excluded), and from with to covers explicit calendar periods: measured totals for accounts and campaigns, explicit gaps and complete=false when coverage is missing, 93 days at most; exact ad set and ad periods are not available here. An ad set, ad group, ad or keyword over dates is read by performance_period. compare=true returns the period and the one just before it in one read. Other non-overlapping periods are separate reads, and rolling-window totals are not independent periods. Meta results can be visits, leads or clicks rather than purchases, so an acquisition CPA needs the conversion definition. Store orders come from store_revenue. Meta figures count each ad set's own attribution setting, as Ads Manager does; a userStatement naming a window ("7-day click only", "1-day click", "without views", "1-day view", "engaged view") counts Meta's purchases and revenue in it, and one asking click versus view ("view-through") adds both windows beside Ads Manager's figure; the result names the window. When the call's userStatement asks what moved on the ads ("what moved this week", "which ads are dying"), it returns the report of what moved instead, at ad level from each Meta account's latest sync: the ads spending more, less, stopped, dying slowly or at a low ROAS, against the period before.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
saas_performanceSaas performanceARead-onlyIdempotentInspect
Stripe billing, RevenueCat revenue metrics and PostHog product performance for SaaS: active subscriptions, trials, normalized MRR, ARR, collected payments, refunds and daily product events. Also returns customer and subscription counts by actual billing cadence when customer-level subscription data is available: how many customers pay annually or monthly. RevenueCat ARR is normalized MRR multiplied by 12. Each source and currency stays explicit, and no attribution is inferred without an identity join.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
site_briefSite briefARead-onlyIdempotentInspect
Reads a public website or store for the requested ads: the page and, for a Shopify store, its public product list. Returns the brand, what the site sells, its language and currency, the first products with their prices, short value propositions and main links; the launch_brief values this settles (destinationUrl, purpose, copy=generate, a proposed campaignName, objective Sales when products were found, the declared language); and for a Google Ads account, suggested Search keywords, headlines (30 characters at most) and descriptions (90 at most) for google_ads_build. The account, budget, countries, dates, Page and pixel are not inferred from the site. Read only.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
store_customersStore customersBRead-onlyIdempotentInspect
Customers imported from the selected Stripe or commerce connection, newest first: name, email domain, country, creation date, order count and imported lifetime spend. Full email addresses are not stored. offset opens the next page when the total exceeds limit.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
store_ordersStore ordersARead-onlyIdempotentInspect
Orders one by one, most recent first: totals, financial status, fulfilment status and line count, for a specific order or status. Totals over a period come from store_revenue in one call.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
store_productsStore productsARead-onlyIdempotentInspect
Up to 50 products from the connected commerce stores, sorted by title: title, price range, unit cost, currency, inventory, status, type, vendor, handle and provider. Individual transactions are in store_orders, aggregated sales in store_revenue.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
store_revenueStore revenueARead-onlyIdempotentInspect
Revenue for the connected stores: gross, refunded and net sales kept apart, with order count, gross profit, break-even ROAS and customer value, each in the store's own currency without conversion. Ad spend is not included; reporting_overview returns it.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
targeting_optionsTargeting optionsARead-onlyIdempotentInspect
Looks up actual platform audience, interest, language or location IDs. A Meta custom audience comes with its kind, Meta's size estimate, retention, state and creation date. Returned candidates are not confirmed launch choices.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
workflows_listWorkflows listARead-onlyIdempotentInspect
Lists up to 100 workspace automations with their IDs, names, purpose (why each exists, in its requester's words), enabled state, definitions and next scheduled run, plus available workflow templates. after takes nextCursor for the next page. Read only; workflows_preview inspects a proposed run without delivery, and automation_propose sets up a new automation, saved off.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
workflows_previewWorkflows previewBRead-onlyIdempotentInspect
Previews an automation or integration source without delivering messages or applying its proposed changes.
Inputs: input (required), accountIds, userStatement, idempotencyKey.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Run Facebook, Instagram and Google ads from Claude, ChatGPT, Grok or Cursor, in plain English.
Run Google Ads and Meta Ads from ChatGPT or Claude: audit wasted spend, create and manage campaigns.
Create, launch, and manage Meta + Google ads from Claude and ChatGPT.
Manage Google, Meta, TikTok, Snapchat and LinkedIn ads, plus GA4, Tag Manager and Search Console, by chat from Claude or ChatGPT. Spend increases need your confirmation first.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceCreate, launch, and manage Meta + Google ads from Claude and ChatGPT. Analytics, strategy, and autopilot automation for any store.1Apache 2.0
- AlicenseNot gradedqualityBmaintenanceConnects Meta (Facebook/Instagram) ad accounts and a Shopify store to AI assistants, letting them read live campaign performance and profit after returns and, with owner approval and a 7-day undo, pause or enable ads, adjust budgets, and build new campaigns that start paused.1MIT
- FlicenseNot gradedqualityCmaintenanceCreate and manage Facebook & Instagram (Meta) ad campaigns from Claude, ChatGPT & Cursor — bulk creation, multi-account, and spend guardrails. Hosted, Meta Verified Tech Provider.3-
- FlicenseNot gradedqualityCmaintenanceConnects AI assistants to Google Ads, Meta Ads, ChatGPT Ads, TikTok Ads and LinkedIn Ads so users can query live campaign performance, budgets and audiences in the account's own currency and prepare changes as proposals that only take effect after explicit approval. Reads run immediately, while every write is staged as an approvable proposal with before/after values, and newly created campaigns, ad sets and ads start paused.1-
Glama MCP Gateway
Add one secure layer between your agents and this server.