The flagged Sentinel files implement an opt-in local security guard and summary reporting for Atris's dashboard. The credential and external-host matches are guard test samples, not code that sends credentials to those hosts.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
Users explicitly run sentinel commands; guard installation is not performed automatically by the package entry point.
Impact
No unauthorized credential transfer, account action, persistence, or control-surface change was identified.
Mechanism
The guard checks agent actions locally. Sentinel event reporting sends fixed, validated summaries to the Atris dashboard.
Evidence
.claude/hooks/sentinel-guard.js.claude/settings.json~/.claude/hooks/sentinel-guard.js~/.claude/settings.json
Network endpoints1
Atris dashboard endpoint: /api/sentinel/events