The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/query.sh utility wraps command-line database clients (sqlite3, psql, mysql) to execute queries. The script uses shell interpolation for the $query and $DB_CONNECTION variables, which presents a risk of command injection if the AI agent constructs these values using untrusted or unvalidated input.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources such as SQL databases, CSV files, and Excel spreadsheets. This creates a vulnerability where malicious instructions could be embedded in the data being analyzed (e.g., in table rows or spreadsheet cells).
Ingestion points: scripts/query.sh (database results) and scripts/analyze_template.py (CSV/Excel loading).
Boundary markers: None identified; instructions do not advise the agent to treat data as untrusted.
Capability inventory: Shell execution (query.sh), Python execution (analyze_template.py), and file system writes.
Sanitization: None; the skill lacks filtering or escaping for external content before processing.
[DYNAMIC_EXECUTION]: The scripts/data-init.sh script dynamically generates a Python script (analyze_template.py) and uses chmod +x to make it executable. While this is part of the workspace setup process, the runtime generation of executable code is a pattern that can be abused to execute arbitrary logic.

data-analyst