loopany
Warn
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
loopany upcommand to start a detached background daemon that persists across user sessions to manage task scheduling. - [REMOTE_CODE_EXECUTION]: The agent is instructed to author, test, and install JavaScript 'workflows' (via
loopany set-workflow) that are executed in a Node.js environment with access to file system and network built-ins. - [DATA_EXFILTRATION]: Configuration details, run milestones, and task results are sent to a remote server URL provided by the user during the initial bootstrap phase.
- [PROMPT_INJECTION]: The skill is designed to process untrusted data from multiple external sources, such as Google Search Console, Reddit threads, and GitHub pull requests. This constitutes a surface for indirect prompt injection that could lead the agent to misinterpret data as instructions or generate malicious code in its automated workflows.
Audit Metadata