The Agent Skills Directory

Command Execution: The skill utilizes various system and AWS CLI commands to manage infrastructure and session states. These include aws dms for migration resources, aws iam for role management, aws secretsmanager for credential handling, and aws s3 for artifact storage. It also uses chmod 600 to secure downloaded SQL scripts and openssl for generating placeholder passwords. These operations are essential to the skill's purpose as a cloud management tool.
Indirect Prompt Injection Surface: The skill processes external data, specifically DDL scripts and conversion assessment reports (CSV) stored in S3. This creates a surface for indirect prompt injection if the ingested content contains malicious instructions. However, the skill implements boundary markers by instructing the agent to wrap generated SQL in specific comments (-- [GenAI-generated]) and requires customer confirmation before applying fixes, which mitigates the risk.
High-Privilege IAM Management: The setup wizard guides the creation of IAM roles and policies. While these involve high-privilege operations, the skill follows AWS security best practices by including aws:SourceAccount and aws:SourceArn condition keys in trust policies to prevent confused-deputy attacks and scoping S3/Secrets Manager policies to specific resources.
Data Handling: The skill manages sensitive data such as database schemas and credentials. It emphasizes the use of AWS Secrets Manager rather than hardcoding credentials and recommends the use of SSL for database connections. Artifacts exported to S3 are handled using standard encryption (SSE-S3).

dms-schema-conversion