USMetrics

[Skill Scanner] Backtick command substitution detected No clear malicious code is present in the provided skill manifest: requested APIs, filesystem reads/writes, and local tool execution are consistent with the stated purpose. However, there are notable security concerns: (1) the skill mandates an immediate background curl POST to http://localhost:8888/notify before any other action — an unusual, coercive local network call that could be abused by a local listener to record invocation metadata or trigger other behavior; (2) the skill relies on executing TypeScript tools shipped in the skill directory via bun, which gives those files permission to perform arbitrary IO and network operations (if the tool files are malicious or compromised they could exfiltrate API keys or system data); (3) loading optional user customization files from a local path that 'override default behavior' increases risk if those files are untrusted. Recommended mitigations: do not run the mandatory curl without inspecting what service listens on localhost:8888; audit the tools/*.ts scripts before executing; run these scripts in a constrained environment; and avoid placing sensitive API keys in an environment accessible to untrusted code. Verdict: SUSPICIOUS but not overtly malicious based on provided content. LLM verification: This skill's documented purpose (US economic metrics collection and analysis) is plausible and some requested permissions (FRED/EIA API keys, read/write data files) are appropriate. However, two elements are suspicious: the hard requirement to run a silent background curl POST to http://localhost:8888/notify before any action, and the instruction to load and apply arbitrary user customization files that override behavior. Those introduce significant risk because they enable local coordination an