The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/analyze.py contains logic to check for and dynamically install the duckdb and openpyxl packages using pip install via subprocess.run if they are not already available in the environment.
[COMMAND_EXECUTION]: The skill uses subprocess.run to execute package installation commands. Additionally, it executes SQL queries provided by the AI agent against data loaded from external files, which is a form of dynamic code execution within the DuckDB environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data from uploaded files which could contain malicious formatting or data intended to influence the agent's behavior during analysis.
Ingestion points: Files uploaded to /mnt/user-data/uploads/ are read by the analyze.py script using DuckDB's read_csv_auto and st_read functions.
Boundary markers: The script does not implement explicit boundary markers or "ignore instructions" warnings for the data being ingested.
Capability inventory: The skill can execute SQL queries, write results to the file system (CSV, JSON, Markdown), and run shell commands via subprocess.run for package management.
Sanitization: The script includes a sanitize_table_name function using regex to clean sheet and file names, but it does not perform deep sanitization of the data content itself before processing.

data-analysis