The Agent Skills Directory

[SAFE]: The skill performs directory listings and metadata scans of project-specific files (e.g., .kiro/specs/, .kiro/steering/) to determine the project state. These operations are limited in scope and do not involve sensitive system paths.
[SAFE]: The skill uses sub-agents for codebase exploration and research, which is a common pattern for managing context budget in AI agent environments. It explicitly limits findings to under 200 lines to prevent context overflow.
[SAFE]: The file writing operations are restricted to the project's own .kiro configuration directory for persistence of discovery results. This is standard behavior for a project management skill.
[SAFE]: No network exfiltration, hardcoded credentials, or obfuscation techniques were detected in the instructions or configuration files.

kiro-discovery