The Agent Skills Directory

[SAFE]: The skill implements strong security boundaries by explicitly forbidding the agent from outputting or echoing real secret values (e.g., '真实值必须脱敏，不写入报告或终端输出'). It focuses on variable names and metadata rather than content.
[DATA_EXPOSURE]: While the skill interacts with sensitive configuration files (.env), its primary purpose is to improve security posture by identifying hardcoded secrets and ensuring correct .gitignore usage. It lacks network capabilities to exfiltrate any accessed data.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it scans external project files (source code and configuration). However, the risk is mitigated by strict instructions regarding output formatting and the requirement to obtain user confirmation before making any changes.

env-manager