memory-hygiene

[Skill Scanner] Destructive bash command detected (rm -rf, chmod 777) This skill appears functionally consistent with its stated purpose (memory audit/cleanup), and contains no signs of obfuscated malware or network exfiltration. However it carries moderate operational risk: it instructs and automates destructive filesystem operations (rm -rf), encourages storing sensitive locations (credential locations), and adds cron jobs that perform wipes and re-seeding without documented safeguards. Treat as SUSPICIOUS/operationally risky rather than malicious — safe only if run by trusted operators with least-privilege permissions, confirmations for destructive steps, and a policy forbidding storage of actual secrets or credential tokens. LLM verification: This skill's stated purpose (audit/wipe/reseed a LanceDB-based agent memory) matches the actions it documents, but it includes high-impact destructive and data-handling operations with insufficient safety controls. The rm -rf wipe, automated monthly cron that deletes and re-ingests data, and an explicit encouragement to store account/credential locations are the primary risks: they can cause irreversible data loss or consolidate sensitive information into the vector DB, which may be exposed if a