jaan-to-codex-pack

SUSPICIOUS. The core behavior mostly matches a code-integration skill, and the documented package-manager/MSW commands point to legitimate ecosystems rather than attacker infrastructure. However, the skill has unusually broad write/edit authority, can execute unpinned install/run commands, relies on opaque local pre-execution protocol documents, and encourages follow-on skill execution. No confirmed credential theft, covert exfiltration, or malicious pre-execution command is visible, so this is not malware, but it carries meaningful workflow and supply-chain risk.