pushary@1.9.101
Human-in-the-loop (HITL) for AI agents: ask a person for approvals or answers by push notification. One command connects Claude Code, Codex, Cursor, Gemini CLI and more.
AI security review
scanned · by lpm-firewall-aiThe flagged behaviors are explained by Pushary’s agent-control purpose: its key is sent to Pushary’s service, and phone-driven sessions use Claude’s SDK with its permission callback. Setup has no lifecycle hook, and shell alias persistence is opt-in through an interactive confirmation.