browse_testing_ads
ChatGPTBrowse authorized Creative Testing ads across batches by decision, date or keyword. Returns their latest monitoring decision (WINNER, LOSER, PROMISING or CONTINUE), spend, conversion events and parent batch. Results describe test ads and their monitoring state.
can_say_craft_claim
ChatGPTGuardrail for craft-performance claims. Checks whether analytics-engine findings support a claim at the requested scope, and returns safe wording plus caveats.
get_atlas_report
ChatGPTRead a previously granted Ad Atlas report by its reportId, including creative-format mix, diversity percentile, coverage, gaps and an interactive map. Findings and the map come from the latest Atlas map the report has been derived on; they represent associations, not ROAS or causation. This read runs no new compute and retrieves no private ad-account metrics or media.
get_atlas_request_status
ChatGPTRead the existing Atlas request and its processing status for the signed-in member’s verified email. This read starts no analysis, retries, refreshes or charges.
get_batch_craft_readout
ChatGPTRead batch-level craft findings for one Creative Testing batch. The findings apply to that individual test; they do not establish account-wide effects.
get_client_craft_playbook
ChatGPTRead account-level craft findings for one authorized organization by orgSlug. Returns associations between craft patterns and performance, with confidence, support, caveats and freshness metadata.
get_creative_testing_batch
ChatGPTRead one authorized Creative Testing batch by batchId. Returns status, platform, campaign and adset IDs, budget, linked batches, and each ad’s latest monitoring decision with spend, conversion events, CPC z-score and p-value. Includes the stored seven-day report when includeSevenDayReport is true and a report exists.
get_framework_access
ChatGPTRead the signed-in account’s access to Atlas reports and connected-account analytics. Returns capability information and the reasons features may be unavailable.
get_test_ad_scale_performance
ChatGPTFind scaled platform ads associated with an authorized Creative Testing ad through perceptual-hash matching. Returns platform ad IDs, campaign and adset context, similarity scores and whether each match originated in the scale-campaign auto-launch pipeline.
list_atlas_reports
ChatGPTList up to 20 recent Ad Atlas reports granted to your verified sign-in email, without requiring a paid Framework plan. Reports contain existing creative research, not live performance. The date listed is when access was granted; reading a report returns the latest map it has been derived on. No imports, new analysis, emails or report refreshes are started.
list_creative_testing_batches
ChatGPTList authorized NewForm Creative Testing batches with lifecycle-status, name and tag filters and offset pagination. Returns id, name, status, platform, creation date, ad count, batch tags as { category, tag, parentTag }, totalCount and tallies of each ad’s latest monitoring decision: WINNER, LOSER, PROMISING or CONTINUE.
list_metrics
ChatGPTList the available metric vocabulary for one organization: cross-platform defaults, metrics for its connected platforms and saved custom metrics. Returns each metric’s name, id, description and type.
list_organizations
ChatGPTList the NewForm organizations available to the signed-in account for connected-account analytics, including each organization’s slug and connected ad platforms. Returns an empty list when no organization meets the account’s entitlement and permission requirements.
query_ad_metrics
ChatGPTQuery aggregated advertising performance across Meta, TikTok, Google and Snapchat for one organization. Supports account, campaign, adset, ad, platform and total levels, time ranges and grains, metric names or IDs, audience and placement dimensions, filters, sorting and pagination. Returns rows with the requested dimensions and metrics. Money uses the organization’s preferred currency, with USD as the fallback when none is configured or organizations in scope disagree; the currency field identifies it. Meta and TikTok date presets use each ad account’s reporting timezone. Answers come from the nightly stored ad history when it reproduces the query exactly over finished days, and from the live platform reporting APIs otherwise (a live query can take several seconds); the source field says which, and as of when.
request_atlas_report
ChatGPTRequest one free creative-research Atlas for the brand associated with the signed-in member’s verified company email. With confirm=false, returns a preview of the request, privacy link, public-library ad processing and report-status emails. Submission requires the user’s explicit confirmation and privacy acceptance, represented by confirm=true and privacyAccepted=true. No paid plan is required. Company-domain and global quotas apply. Retries reuse the original request. Report refreshes, ad launching, budget changes and arbitrary prompts are unavailable.
browse_testing_ads
Claudecan_say_craft_claim
Claudeget_atlas_report
Claudeget_atlas_request_status
Claudeget_batch_craft_readout
Claudeget_client_craft_playbook
Claudeget_creative_testing_batch
Claudeget_framework_access
Claudeget_test_ad_scale_performance
Claudelist_atlas_reports
Claudelist_creative_testing_batches
Claudelist_metrics
Claudelist_organizations
Claudequery_ad_metrics
Clauderequest_atlas_report
Claude