Security audit
Security checks for vulnerabilities and agentic risk
This is a local Chinese study-error analysis skill that saves mistake records locally, with no network access or privilege-seeking behavior found.
Install only if you are comfortable with a Chinese-language study tool that saves mistake records locally. Avoid entering sensitive personal information in questions or answers, and review or delete artifact/data/errors.json when you no longer need the saved records. The extra scripts/script.sh is not part of the documented workflow and should not be run unless you specifically want its local logging/datastore behavior.
The description and body present the skill primarily in Chinese while also including some English, but there is no statement that users can choose their preferred language. This can violate a language/locale policy when a skill appears to impose a default language without explicit opt-in.
The natural-language interface, help text, and instructional output are all fixed in Chinese, which forces a specific language experience on users. There is no indication that the skill is region-specific or that users can choose another language.
The script stores user-provided question text, wrong answers, and correct answers into a local JSON file without any explicit notice, consent prompt, retention policy, or masking. In an educational context, this can expose potentially sensitive study content or personal data entered by users to other local users, backups, logs, or later unintended reuse.
The header and help output present the tool as "error-analysis" and describe it as a utility in that context, but the implemented functionality only stores, lists, searches, and exports arbitrary log entries. This is not just incomplete documentation: the documented identity of the tool suggests an analysis-focused purpose that the code does not implement.
All command arguments are written to a history file, which can silently persist sensitive data such as search terms, tokens, filenames, or operational details. In a skill context, this is more dangerous because users may pass incident data or secrets assuming the tool is ephemeral, creating unintended local data retention and possible later disclosure to other local users or processes.
The add command stores arbitrary user-provided content to disk without clear disclosure, which can capture sensitive notes, credentials, or incident details the user did not expect to persist. Because this tool presents itself as an analysis utility rather than a datastore, the mismatch increases the likelihood of users entering confidential information that is then retained locally.
The file’s instructional content is written in Chinese and does not indicate that users may choose another language or locale. Under the policy for natural-language violations, forcing a specific language without opt-in is reportable unless the constraint is clearly justified.
No suspicious patterns detected.
