diff-scanning-with-aws-security-agent
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The skill executes standard system commands including
gitto identify code changes,zipto prepare files for analysis, and theawsCLI to interact with cloud services. These operations are restricted to the local workspace and the user's AWS environment. - Data Transfer: Source code and diff patches are uploaded to an Amazon S3 bucket within the user's own AWS account. This is a standard architectural pattern for cloud-based security analysis where the processing occurs on service-managed infrastructure.
- Indirect Prompt Injection: The skill reads findings generated by the security agent service. Because these findings describe content from the analyzed source code, they represent a potential surface for indirect instructions.
- Ingestion points: Results from
aws securityagentcalls are retrieved and processed as described inSKILL.md. - Boundary markers: The skill does not explicitly define delimiters or boundary markers for the service output.
- Capability inventory: The skill environment includes shell command execution and network access via the
awsCLI. - Sanitization: Results are reported to the user or written to a local file without explicit sanitization or filtering steps mentioned in the skill instructions.
Audit Metadata