edge-hint-extractor

SUSPICIOUS. The stated purpose is coherent for local market-data transformation, and the default/local-file workflow looks benign. Risk comes from the optional arbitrary external LLM command: it is underdocumented, unverified, and can route input data to unknown third-party code, making the skill moderately risky even without direct credential handling.

This module is primarily a local data-processing tool with safe parsing practices (json.loads and yaml.safe_load). However, it includes a high-impact optional feature: it can execute an arbitrary external command provided by the user via --llm-ideas-cmd (subprocess.run) and then parse the resulting stdout YAML to generate output. If an attacker can influence CLI arguments or execution context, this becomes an effective arbitrary code execution vector. Aside from the external-command capability and caller-controlled output paths, there is no clear evidence of embedded malware, obfuscation, network exfiltration, or credential theft within this module.