The Agent Skills Directory

[SAFE]: The skill instructions enforce a defensive security posture by explicitly requiring that investigations remain read-only. It prohibits the installation of software, modification of system state, or refreshing of credentials during exploration.
[SAFE]: No evidence of prompt injection, data exfiltration, or obfuscation was detected. The instructions prioritize setting a 'budget' (time or probe count) to prevent unbounded execution and resource exhaustion.
[SAFE]: The skill uses local file storage for large intermediate datasets collected during investigation, which is a recommended practice to avoid exposing large volumes of raw data directly in the prompt context.
[SAFE]: Analysis of the evaluation fixtures and package configurations revealed no hardcoded secrets, malicious dependencies, or suspicious network patterns. All external references are limited to mock organizational packages for testing purposes.

bound-the-unknown