@funny-booth/agent-core@0.1.58
AIエージェント起動CLI(共有ランチャー + MCPサーバー)
AI security review
scanned · by lpm-firewall-aiThe package intentionally obfuscates its published runtime code, and the launcher includes a credential-bearing request whose destination and trigger could not be confirmed from the inspected excerpts. This supports a warning for unresolved behavior, but does not establish a concrete attack.