storing-and-querying-vectors
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- AWS CLI Command Execution: The skill instructs the agent to execute various
aws s3vectorsandaws bedrock-runtimecommands. These are used for administrative and data operations within the AWS environment, such as creating buckets, managing indexes, and generating embeddings. - Indirect Prompt Injection Surface: The skill processes external data, such as text input for embeddings and metadata associated with vectors. This data enters the agent's context and could theoretically contain instructions intended to influence the agent's behavior, which is a common consideration for skills interacting with databases or external APIs.
- Local File Access: The instructions describe reading from a local file (
invoke-model-output.json) to process output from the Bedrock service. This is a standard method for handling CLI tool responses and extracting specific data points like embeddings.
Audit Metadata