parallel-monitor
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides setup instructions that involve downloading and piping an installation script from the vendor's official domain (
https://parallel.ai/install.sh) to bash. This is a common pattern for installing developer tools from verified author sources. - [COMMAND_EXECUTION]: The skill uses the
Bashtool to runparallel-clicommands. These commands manage web monitors, including creating, listing, updating, and deleting monitoring tasks as requested by the user. - [PROMPT_INJECTION]: The skill features a surface for indirect prompt injection or command injection by interpolating user-provided text into shell command arguments.
- Ingestion points: The user's search or monitoring query is used as the
<query>argument in theparallel-cli monitor createcommand inSKILL.md. - Boundary markers: The instructions wrap the user query in double quotes in the command template.
- Capability inventory: The skill is authorized to use the
Bashtool to executeparallel-clioperations. - Sanitization: There are no specific instructions for the agent to escape or sanitize shell metacharacters (such as backticks or dollar-sign expressions) within the user's query before execution, which represents a potential injection surface.
Audit Metadata