The Agent Skills Directory

[COMMAND_EXECUTION]: The skill specifies a hardcoded absolute path to a Java binary at /Users/elliottleonard/Documents/Cursor/CLI/domoutil.jar. Executing binaries from user-specific home directories is a security risk as it bypasses standard environment configurations and creates a dependency on a path that could be targeted for local privilege escalation or binary hijacking.
[COMMAND_EXECUTION]: The provided Python and Shell scripts dynamically assemble multi-line command strings by concatenating user-provided or environment-derived data (e.g., friendly_name = csv_file.replace(...)) and piping them to a subprocess. This pattern is vulnerable to command injection if the input data (such as filenames in a processed directory) is maliciously crafted.
[PROMPT_INJECTION]: The skill implements an indirect prompt injection surface by consuming untrusted external CSV data to automate schema derivation and dataset generation.
Ingestion points: The skill reads external CSV files via derive-schema -d /path/to/data.csv and upload-dataset -f /path/to/data.csv in SKILL.md.
Boundary markers: Absent. There are no delimiters or instructions to the agent to disregard potentially malicious content embedded within the CSV files.
Capability inventory: The skill possesses the capability to execute shell commands via subprocess.run and perform network operations using the Domo API.
Sanitization: Absent. There is no evidence of validation or sanitization of the CSV content before it is processed by the CLI tool or the Python script logic.

data-upload-java-cli