The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): Significant vulnerability to indirect prompt injection via ingested project data.\n
Ingestion points: The pipeline_generator.py and terraform_scaffolder.py scripts ingest data from user-provided project and target paths.\n
Boundary markers: The skill fails to define delimiters or include 'ignore embedded instructions' warnings for external content.\n
Capability inventory: The skill description confirms high-privilege capabilities including 'infrastructure automation' and 'deployment automation' across AWS, GCP, and Azure.\n
Sanitization: No evidence of schema validation, input escaping, or sanitization of the processed configuration files exists.\n- [Unverifiable Dependencies & Remote Code Execution] (MEDIUM): The workflow relies on unverifiable external packages.\n
Evidence: The setup guide instructs users to run npm install and pip install -r requirements.txt without providing the manifest files or pinning versions, posing a supply chain risk.\n- [Command Execution] (MEDIUM): The skill executes multiple local scripts that perform system-level tasks.\n
Evidence: Execution of pipeline_generator.py, terraform_scaffolder.py, and deployment_manager.py via python3 to automate DevOps workflows.

senior-devops