The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, such as production application code, IaC templates, and Docker configurations. This creates a surface for indirect prompt injection where malicious instructions hidden in the analyzed content (e.g., code comments or documentation) could attempt to influence the agent's behavior during the audit process.
Ingestion points: Audited project files including source code (**/config/, **/auth/), deployment files, and environment variables.
Boundary markers: The workflow does not explicitly instruct the agent to use delimiters or ignore instructions embedded within the analyzed data.
Capability inventory: The skill utilizes file discovery, searching, and code tracing capabilities to evaluate security risks.
Sanitization: No specific sanitization or input validation for the analyzed content is mentioned.
[SAFE]: The skill consists entirely of instructional Markdown and does not include any executable scripts, binary files, or installation commands.
[SAFE]: The credentials provided in the references/examples.md file (e.g., 'dev-secret-key-123', 'admin123', 'sk_tes...') are clearly documented as illustrative, non-sensitive dummy values used for educational purposes to demonstrate vulnerable patterns.

insecure-defaults