remote-control
Audited by Socket on Aug 26, 2026
2 alerts found:
AnomalySecuritySUSPICIOUS: the skill’s core behavior is coherent with remote mobile access, and the Anthropic remote-control component is official, but it grants unusually broad multi-project bash and file access through a custom unverifiable local MCP server. Main risk is high-impact remote control and data exposure across all registered projects rather than clear malicious exfiltration.
SUSPICIOUS: the stated purpose is coherent, and the remote-control flow aligns with official Claude Code behavior, but the key install and control surface relies on an unverified local script and an unverifiable MCP server. Broad cross-project shell/file access is proportionate to orchestration, yet the unresolved provenance keeps risk high enough to avoid a benign classification.