move-auditor

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill contains deliberate orchestration to collect local source files into bundles and hand them to parallel "Agent" processes (including a DEEP-mode "opus" agent) plus explicit network calls (curl), which creates a high risk of sensitive-data exfiltration to external LLM/agent endpoints — there is no explicit recipient whitelisting, consent/taint checks, or protections for secrets; I found no obvious hidden eval/reverse-shell or credential-harvesting code, but the bundling+agent-network pattern is a high-risk data-exfiltration/backdoor vector.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow reads outsider-authored free text from the local references/attack-vectors/attack-vectors-1.md (and other attack-vectors-*.md plus hacking-agents/*.md) into the LLM context via the bundled agent-*-bundle.md files, and those reference markdown files are not authored by the operating user.