The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill instructs the agent to execute PowerShell scripts using the -ExecutionPolicy Bypass flag. This command explicitly circumvents the security settings of the host operating system that are designed to prevent the execution of unauthorized or untrusted scripts.
Evidence: Multiple occurrences in SKILL.md where powershell -ExecutionPolicy Bypass is used to call scripts in the .\scripts\ directory.
[COMMAND_EXECUTION]: The skill utilizes command-line interfaces to run MATLAB code and system scripts, providing the agent with the ability to execute arbitrary code within the MATLAB environment and shell.
Evidence: Use of matlab -batch for direct execution and PowerShell for wrapper script execution as defined in the 'Preferred Wrapper' and 'MATLAB Discovery' sections of SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from external sources, such as research paper URLs and dataset URLs, which can be exploited to inject malicious instructions into the agent's context.
Ingestion points: The 'Paper Reproduction Mode' requires the agent to record and process content from external 'paper URL' and 'dataset URL' fields (SKILL.md).
Boundary markers: No boundary markers or protective delimiters are prescribed to isolate the untrusted external content from the agent's instructions.
Capability inventory: The skill has access to powerful execution capabilities, including shell access via PowerShell and the ability to run MATLAB scripts (SKILL.md).
Sanitization: There are no instructions for sanitizing or validating the content retrieved from these external URLs before it is processed by the agent.

matlab-runner