The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted third-party source code during security reviews, which creates a potential vector for indirect prompt injection attacks.
Ingestion points: Instructions in SKILL.md (Workflow Process and Core Mission) direct the agent to review external code changes and architectural designs.
Boundary markers: The instructions do not specify the use of delimiters or protective headers to isolate untrusted code content from the agent's instructions.
Capability inventory: The skill provides a Python script utilizing subprocess.run to interact with the system shell for auditing purposes.
Sanitization: There are no explicit requirements or methods defined for sanitizing code content before it is processed by the model.
[DYNAMIC_EXECUTION]: The skill provides a functional Python script for dependency auditing that relies on runtime execution of external commands.
Evidence: The DependencyScanner class in SKILL.md uses subprocess.run to call external CLI tools.
[COMMAND_EXECUTION]: The skill automates the execution of system-level commands through a provided utility script.
Evidence: The Python script in SKILL.md executes npm audit and pip-audit via subprocess calls to analyze project dependencies.

Application Security Engineer