The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill is designed to analyze and process untrusted external content (application code, Terraform files, CI/CD configs) while possessing high-privilege execution capabilities.
Ingestion points: project-path and target-path parameters in scripts, plus general analysis of the working directory.
Boundary markers: None identified in the provided documentation or script usage instructions.
Capability inventory: Capability to run kubectl apply, docker build, npm install, and arbitrary local Python scripts that interact with cloud providers (AWS, GCP, Azure).
Sanitization: No evidence of input sanitization or validation of the files being processed.
Command Execution (MEDIUM): The skill relies on several local Python scripts (scripts/pipeline_generator.py, scripts/terraform_scaffolder.py, scripts/deployment_manager.py) to perform core functions. Without the source code for these scripts, their behavior regarding system access and credential handling remains unverified.
Unverifiable Dependencies (MEDIUM): The development workflow encourages npm install and pip install -r requirements.txt. These commands download and execute code from external registries. Without pinned versions or integrity hashes in the instructions, this presents a supply chain risk.

senior-devops