Security audit
Security checks for vulnerabilities and agentic risk
This music helper uses network lookups and local JSON storage in ways that fit its stated music search, lyric, cache, favorites, and playlist features.
Install only if you are comfortable with music searches and lyric/ranking lookups being sent to Netease music endpoints and with favorites, playlists, play counts, and cache data being stored as local JSON files. Review or clear those local files if music preferences are sensitive on your device.
SQP-3 applies to all file types and covers natural-language policy violations such as forcing a specific language. This skill's documentation, examples, and interaction patterns are entirely in Chinese, with no indication that users can choose another language or that the language constraint is an intentional region-specific limitation.
The skill requests or implies network and file read/write capabilities in its markdown, but it does not declare an explicit tool scope such as permissions or allowed-tools. This creates a trust and containment gap: the runtime may permit broader access than users or reviewers expect, increasing the risk of unintended file access or external data transmission.
The title, description, trigger rules, examples, and interaction patterns are entirely in Chinese, and the skill does not indicate that users may choose another language. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale restriction is explicitly justified.
The trigger phrases are very broad and match ordinary music-related conversation, so the skill may activate when the user did not intend to use it. Unintended invocation can cause unexpected network calls, retrieval of external content, or local state changes such as playlist writes.
The skill stores favorites and playlists on local disk, including user preference history such as saved songs and play counts, but it does not clearly notify users of this behavior. Silent local persistence can expose personal preference data to other local users, backups, or unrelated processes and may violate user expectations around privacy.
The skill indicates it will contact external music and lyrics services, but it does not warn users that their queries, song interests, or lyric fragments may be transmitted to third parties. Music queries can reveal language, mood, interests, or other behavioral signals, so undisclosed sharing creates a meaningful privacy risk.
The description, trigger examples, supported moods/scenes, and response strings are all in Chinese, with no indication that other languages are supported or that the user can choose a locale. Under the policy, forcing a specific language without opt-in is a natural-language policy violation.
The Accept-Language header is fixed to prefer zh-CN/zh, which imposes a specific locale behavior in the skill. The file also presents its user-facing descriptions and outputs in Chinese, but does not document any opt-in or configurable language selection.
The markdown advertises real-time lyric lookup and later documents Netease API/crawler integration and optional self-hosted API deployment, which implies external network requests. Under SQP-2 for markdown files, the description should warn about privacy-relevant behavior when user inputs may be sent to third-party services, but no such warning is present.
This is a markdown file, so SQP-2 applies to omitted warnings about behaviors affecting user data or system state. The README notes that favorites.json and playlists.json are automatically generated and promotes 收藏管理/歌单管理 features, but it never explicitly warns users that invoking these features will create and persist local data files.
This code persists user music preferences to favorites.json and playlists.json on disk, which affects user data storage. The file write happens silently with only an internal docstring, and there is no visible confirmation prompt or privacy/storage warning in this file's user-facing help text.
No suspicious patterns detected.
