script-generator
Fail
Audited by Gen Agent Trust Hub on Apr 20, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to generate executable scripts and provide an output field named
EXECUTIONcontaining the exact command string (e.g.,python3 OUTPUT_DIR/artifacts/task_name.py) to run them. This facilitates the execution of dynamically generated code on the host system. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests untrusted data via the
TASK,TARGETS, andCREDENTIALSfields inSKILL.mdto construct script logic. There are no boundary markers or sanitization procedures defined to prevent malicious tasks from altering the script's behavior. The capability inventory includes the generation of Python, PowerShell, and Bash scripts with access to network libraries. - [DATA_EXFILTRATION]: The skill defines a protocol for handling raw credentials, including passwords and hashes, alongside external network targets. This combination, when processed by a script generator without oversight, enables the creation of scripts that can exfiltrate sensitive authentication data to arbitrary external endpoints.
Recommendations
- AI detected serious security threats
Audit Metadata