The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The example command for CreateDBInstance in SKILL.md includes a hardcoded password Secret1234!. While likely intended as a placeholder, providing hardcoded secrets in instructions can lead to the deployment of insecure or default credentials.
[COMMAND_EXECUTION]: The RDSTool provides extensive lifecycle management capabilities including CreateDBInstance, ModifyDBInstance, and DeleteDBInstance. These commands allow the agent to perform high-impact modifications to cloud infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions designed to influence the agent behavior. 1. Ingestion points: The agent retrieves potentially attacker-controlled content via DownloadDBLogFilePortion (database logs) and DescribeDBInstances (resource tags and identifiers) as defined in SKILL.md. 2. Boundary markers: The instructions do not define delimiters or warnings to treat retrieved log content or metadata as untrusted data. 3. Capability inventory: The agent has the ability to perform significant infrastructure changes (Create, Modify, Delete) using the RDSTool. 4. Sanitization: The skill does not provide mechanisms for sanitizing or filtering input from database logs or RDS resource attributes before processing.

awsflow-rds