The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains automated Python logic in sub-skills/implementation-playbook.md that interacts with cloud service APIs (e.g., AWS Boto3). This code has the capability to perform administrative actions such as stopping/starting EC2 instances, modifying instance types, and altering Lambda function concurrency for the purpose of cost management.
[DATA_EXFILTRATION]: The skill's primary function involves reading sensitive financial and infrastructure data from cloud billing APIs (e.g., AWS Cost Explorer) and CloudWatch metrics. This data is collected and processed locally to generate optimization reports and dashboards.
[EXTERNAL_DOWNLOADS]: The dashboard generation templates in sub-skills/implementation-playbook.md reference well-known and trusted CDN providers (jsDelivr and d3js.org) to load standard JavaScript libraries for data visualization.
[PROMPT_INJECTION]: The skill's logic in sub-skills/implementation-playbook.md processes untrusted data from the cloud environment, specifically resource tags, resource names, and CloudTrail event details, which could be influenced by an attacker to manipulate tagging or cost allocation automation.
Ingestion points: Ingestion occurs in sub-skills/implementation-playbook.md where resource metadata and CloudTrail logs are retrieved via API calls.
Boundary markers: No explicit markers or instructions are provided to the agent to distinguish between system instructions and data embedded within resource metadata.
Capability inventory: The skill possesses significant capabilities in sub-skills/implementation-playbook.md, including stop_instances, modify_instance_attribute, put_bucket_lifecycle_configuration, and put_function_concurrency.
Sanitization: The provided code samples do not include explicit sanitization or validation logic for the metadata read from external sources.

database-cloud-optimization-cost-optimize