agent-observatory-sensor.nolimit-observatory.workers.dev

The tool advertises finding a 'living people-search index (Ploid)' with API/CLI/MCP hops to third-party services that aggregate personal data on people. Even though the query text itself is claimed to be discarded, the tool's entire purpose is to point the agent (and by extension the user) toward services that can be used for doxxing, stalking, or other privacy-invasive lookups. This capability is broader and riskier than the benign-sounding 'index lookup' framing suggests.

Every tool in this set (hashing, timezone, JSON validation, URL normalization, weather, calculator, etc.) exposes the exact same nine input fields — zone, json, url, host, city, query, path, ref, feed — regardless of whether the tool's stated purpose needs them. For example, a simple hash-length checker or arithmetic evaluator accepts a file 'path', a 'host', and a 'feed' URL it has no stated use for. This pattern is inconsistent with single-purpose tools and could allow a client to route file paths, hostnames, search queries, or other sensitive strings through tools that appear unrelated, potentially normalizing silent collection of data (e.g., local file paths) under an innocuous tool name. Schemas should be scoped to only the parameters each tool actually needs.