inquest

SUSPICIOUS. The skill's research purpose is plausible, but its actual footprint is broad and under-scoped: it pushes the agent to query every reachable evidence source, ingest untrusted external content, and rely on separate tools (`deslop`, `muster`) outside the reviewed skill. There is no confirmed malware or direct credential theft in the text, but the combination of broad MCP reach, prompt-injection exposure, and transitive tool trust makes this a medium-risk agent skill.