Security audit
Security checks for vulnerabilities and agentic risk
This is an overview-only coding methodology skill with no executable behavior, though its trigger wording and package slug should be reviewed for clarity.
Before installing, confirm you want a Chinese-language Apollo methodology overview that may activate on common coding requests. The publisher should fix the slug/name mismatch and narrow or clarify activation wording, but the inspected artifact does not show hidden code, credential access, persistence, or destructive actions.
The trigger phrases are very broad (e.g. '开发新功能', '修bug', '写计划', '开始项目') and can match common user requests that do not explicitly ask for this specific skill. That creates prompt-scope creep: the skill may activate unexpectedly and steer unrelated conversations into its workflow, increasing the chance of unintended behavior or policy/workflow override in downstream agent orchestration.
The manifest description is written as a direct user-facing promise entirely in Chinese, with no indication that language selection is optional or tied to a region-specific use case. Under the language/locale policy, forcing a specific language without user opt-in can be a natural-language policy violation.
The manifest slug in _meta.json ('sysflow-overview') does not match the provided skill context name ('apollo-overview'), indicating an identity inconsistency. While this is not directly exploitable code execution, it can enable skill confusion, mislabeling, or accidental substitution during packaging, review, or deployment, which weakens trust and can hide the true provenance or purpose of the skill.
No suspicious patterns detected.
