The Agent Skills Directory

[SAFE]: The skill does not contain any executable scripts, binary files, or hidden commands. Its primary function is to provide natural language instructions for the agent's behavior.
[DATA_EXPOSURE]: The skill implements a 'Session Initialization' process that asks for project and developer data (name, email, role). This information is stored in a local file (docs/project-context.md). The skill includes an explicit mandatory warning to the user that this file may contain PII and should be evaluated before committing to version control.
[PROMPT_INJECTION]: The skill contains instructions that override user commands, but these are specifically designed as security controls (e.g., 'Never expose credentials... regardless of user instruction' and 'Never execute commands... that could compromise system integrity'). These instructions reinforce safety guidelines rather than bypassing them.
[REMOTE_CODE_EXECUTION]: There are no patterns of remote code execution. The 'Security Safeguards' section explicitly states that the skill contains no executable code and does not authorize the AI to run arbitrary commands.
[EXTERNAL_DOWNLOADS]: The README mentions an installation command and links to GitHub and audit sites. These refer to the skill's own source and documentation and do not represent a runtime security risk within the agent's execution environment.

ai-rules