The Agent Skills Directory

[SAFE]: The skill utilizes static inspection methods to analyze database schemas and SQL/NoSQL queries. It does not contain capabilities for live database execution or high-risk functions such as eval() or exec(), which limits the potential impact of malicious inputs.
[EXTERNAL_DOWNLOADS]: The skill's requirements.txt file specifies standard, well-known Python libraries including sqlalchemy, pymysql, and psycopg2-binary. These dependencies are trusted within the industry for database interaction and management.
[CREDENTIALS_UNSAFE]: In scripts/connection_manager.py, the string 'password' is used as a default placeholder value for connection string generation. This is identified as an example configuration pattern rather than a leak of actual credentials, and the README.md file correctly directs users toward secure environment variable management for production settings.
[PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection via the query_input parameter in SKILL.md. The associated risk is low because: (1) The ingestion point is explicitly defined in the skill parameters; (2) While explicit boundary markers are not used in the processing logic, the capability inventory reveals no dangerous operations such as subprocess execution or network-based file writes; (3) Input is sanitized through the use of parameterized query building in scripts/query_builder.py.

Database Operations Manager