The Agent Skills Directory

[SAFE]: The skill instructions define a benign architectural design workflow. It focuses on documentation, design patterns, and structured reasoning using internal tools.
[DATA_EXPOSURE]: The skill uses Grep and Glob to analyze the local codebase to understand the existing system. This is standard, non-malicious functionality for a development-focused agent skill and does not involve unauthorized data access or exfiltration.
[PROMPT_INJECTION]: The skill processes data from the local repository via search tools, which technically constitutes an indirect prompt injection surface. However, the risk is negligible as the skill lacks high-risk execution capabilities (like arbitrary shell access or network requests), and its purpose is limited to generating documentation and design artifacts.
Ingestion points: Grep and Glob tool usage described in SKILL.md.
Boundary markers: Not explicitly defined for tool output.
Capability inventory: File writing restricted to ./artifacts/ and scratchpad/ directories.
Sanitization: Not explicitly defined.

architect