The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to dynamically generate and execute a Python script (scripts/export_csv.py) to perform data conversion. This pattern represents a high-capability dynamic execution flow.
[COMMAND_EXECUTION]: The instructions explicitly direct the agent to circumvent platform security filters by avoiding shell heredoc syntax that triggers safety warnings, instead instructing the agent to write executable files directly to the file system to evade detection.
[DATA_EXFILTRATION]: The skill accesses and processes sensitive data files, including investor_contacts.json and final_output.json, which contain personal contact information such as emails and LinkedIn profiles. The combination of dynamic script execution and sensitive data access presents a significant risk surface.
[COMMAND_EXECUTION]: The skill exhibits a vulnerability surface for indirect prompt injection. It ingests data from various JSON files without specified sanitization or boundary markers while maintaining the capability to generate and execute code. Maliciously crafted data within the input files could potentially influence the logic of the generated conversion script.

export-csv