To find and map the digital infrastructure enabling online information operations. But a purpose built tool was needed to gather diverse traces and begin to understand the links between cases, platforms, services, datasets, and reporting. This is the first glimpse of a larger WebMCP online observatory project mapping digital phenomena: its direction remains open, and building the instrument is already part of the discovery.
People usually encounter an information operation as a post, advert, headline, or takedown. This proof of concept helps a researcher trace the less visible services, data, platforms, and infrastructure that may enable it.
WebMCP gives a researcher and an AI agent the same inspectable surface: search, follow, compare, and retain leads while the researcher keeps the judgement.
Start with the visible trace, then test whether related services, records, and infrastructure recur around it. A match opens a research lead; it does not settle intent or responsibility.
Watch the wider field arrive.
A connected agent can publish current, directly linked reporting here and label each lead by information-warfare relevance and ecommerce-stack position.
Ask the agent to refresh this feed
“Find current, directly cited reporting relevant to information warfare and its enabling ecommerce stack, then publish the strongest leads to this page.”
Ask the wider field a question.
Ask a question or choose a starting prompt. The prototype searches its curated records and linked source index, then returns a small candidate set to inspect, retain, and turn into a possible output.
AGENT INSTRUCTIONHow the research brief is composed
Act as a careful research assistant. First use the source catalogue and stack filter when they help narrow the question; then search the linked data collections and current reporting for a focused question about information warfare and the ecommerce stack. Return a small candidate set and state the source URL, access or download note, date or recurrence signal, geography, stack position, evidence status, and why each item is relevant. Keep direct observation, support, inference, and dispute separate. Ask the researcher which records to retain, then offer a mini report, timeline, threshold note, or network map. Do not infer coordination, attribution, or responsibility from shared infrastructure. In this proof of concept, treat external links as starting points: they are not downloaded or independently verified records.