The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Automated security scanners identified confirmed malicious content in several reference files belonging to the skill's DevOps guidance lens. The affected files include references/lenses/devops-guidance/DLCD04.md, DLEAC04.md, DLCD05.md, DLEAC07.md, and DLADS03.md. These files are intended to provide framework guidance but triggered malware reputation alerts from a confirmed reputation scanner.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted data from user codebases, IaC templates, and incident reports to generate findings, creating a significant attack surface.
Ingestion points: Workload descriptions, IaC files, application code, and incident context are ingested in Step 1 and Step 4 of the core instruction files (SKILL.md, SKILL-devops-agent.md).
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present when processing the ingested codebase data.
Capability inventory: The agent can generate machine-readable JSON reports and, if confirmed by the user, create tracking issues in external systems using the gh CLI tool.
Sanitization: There is no evidence of explicit sanitization or filtering of codebase content before it is processed by the AI agents.

aws-well-architected-framework-review