The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): Database interactions are performed through a structured Remote Procedure Call (RPC) interface. This design avoids raw SQL string concatenation and prevents SQL injection vulnerabilities by treating target parameters as data.
[DATA_EXFILTRATION] (SAFE): The skill ensures data isolation by filtering queries with a mandatory user_id. No hardcoded credentials, sensitive file access, or unauthorized network activity was detected.
[REMOTE_CODE_EXECUTION] (SAFE): The skill relies on static logic within Python and SQL functions. It does not employ dynamic code execution functions like eval() or exec(), nor does it attempt to download or execute external scripts.

fuzzy-matching