Nothing here answers a question about a brand in one call — you assemble the set, fetch it, and then the search and aggregation tools have something to work on. That is more setup than a lookup tool, and it is also why comparing two sets or aggregating ads by label produces anything useful. The skills exist to carry a newcomer through that first assembly.
A connector to an Adology workspace, exposing 27 tools across seven groups. Knowledge sets are the unit of work — you add brand, influencer, search or discussion feeds to one, trigger a fetch, then search, aggregate and analyse what came back.
- create_knowledge_set, add_feed and batch_add_feeds — build the tracked universe for a brand and its competitors
- trigger_fetch, list_workflows and get_workflow_status to start a scrape and watch it finish
- discover_brands across TikTok, Instagram, Facebook and YouTube, with get_suggestions for adjacent accounts
- search_items within a set, search_all across the workspace, and content_intelligence_search for semantic plus label-filtered search
- analyze, list_labels and aggregate_items to turn a pile of ads into counts by hook, format or whatever the labels capture
- get_table_data for tabular output, get_items and get_item_detail for the ad itself with its analysis and media
- compare_knowledge_sets to put two tracked sets side by side
- save_to_collection, list_collections and get_collection to keep what you found
- whoami for the account, organisation, team and plan behind the session
- Eleven guided skills covering onboarding, brand building, creative strategy, audience deep dives, influencer coaching and vetting
An Adology account. Sign-in is OAuth 2.1 through Stytch and happens the first time a tool is called, so there is no API token to store. Add the hosted endpoint as a connector, or install it as a plugin from the marketplace inside a Claude Code session. Every tool is authorised against the caller's workspace, and cross-tenant reads are refused at the data layer.
Build from source — clone the repository and build it, then deploy it and point your client at the endpoint