skill-discovery

The file itself does not contain embedded malware, hardcoded credentials, or explicit network exfiltration commands. However, it prescribes agent behavior that materially increases supply-chain and data-exfiltration risk: mandatory invocation of skills with a very low applicability threshold, broad local file-read tooling, mandated memory persistence, and absence of provenance or approval controls. These combine to create a high-probability path for transitive attacks (malicious skill propagation, secret disclosure, persistence of sensitive data). Recommended mitigations before granting execution: require signature/provenance checks for invoked skills, add human-in-the-loop approval for non-verified skills, restrict and audit filesystem access of skill tools, sanitize or encrypt memory writes, and make invocation non-mandatory or raise the applicability threshold. Treat this skill as high-risk for autonomy abuse and supply-chain exposure until controls are implemented.