The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses directory exploration commands like ls and Glob to analyze project structure and identify the existing technology stack.
[PROMPT_INJECTION]: There is a potential surface for indirect prompt injection as the agent reads external data from project files (e.g., package.json, pyproject.toml) and user input to form its architectural recommendations. However, the skill's lack of high-privilege capabilities (like file writing or network access) limits the impact of such an attack.
[DATA_EXPOSURE]: The agent intentionally reads project manifests and configuration files to extract context about the existing development environment, which is necessary for its primary purpose of providing relevant architectural advice.

tech-hub-brainstorm