rlm-execute
Warn
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill enables the execution of arbitrary Python code snippets via the
interp.execute()method. This provides extensive capability to perform complex logic, file system operations within the sandbox, and stateful processing. - [REMOTE_CODE_EXECUTION]: The
DaytonaInterpreterinitialization uses arepo_urlparameter (e.g.,https://github.com/your-org/your-repo), which indicates that the execution environment is populated by cloning and potentially executing code from an external repository. - [EXTERNAL_DOWNLOADS]: The skill facilitates downloading code and environment configurations from remote Git repositories into the sandbox execution environment.
- [COMMAND_EXECUTION]: The skill provides an example of reading local file content (e.g.,
scripts/analysis.py) and executing it within the sandbox, which could lead to arbitrary code execution if the local script path or content is influenced by untrusted data.
Audit Metadata