T09 · Insecure Skill Coding Practices
- Location
scripts/disk_butler.py:86- Finding
Environment-Controlled Cleanup Roots Permit Deletion Outside Trusted Cache and Temporary Directories
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed local disk-cleanup tool, but its deletion safeguards rely on environment-derived Windows paths, which makes it worth reviewing before use.
Install only if you are comfortable with a local cleanup tool that can delete files after explicit confirmation. Run doctor, plan, and clean without --yes first, inspect the listed paths carefully, avoid elevated/admin shells, and do not use it from wrapped or modified environments where USERPROFILE, LOCALAPPDATA, APPDATA, SystemRoot, TEMP, or TMP may be altered.
scripts/disk_butler.py:86Environment-Controlled Cleanup Roots Permit Deletion Outside Trusted Cache and Temporary Directories
The file’s user-facing description and command output are written exclusively in Chinese, and there is no indication that users can opt into another language or locale. Under the policy rule, forcing a specific language without user choice is a natural-language policy violation.
The entire skill description is presented only in Chinese, including usage, safety notes, and warnings, with no indication that other languages are supported or that the skill is intentionally limited to a Chinese-speaking audience. Under the policy rule for language or locale constraints, this can be a natural-language policy issue because users are not given an opt-in or alternative.
The manifest sets locale: zh-CN and states the skill is designed for Chinese users, which imposes a language/locale constraint. The file does not offer a selectable language option or explicit user opt-in within the skill, so this is a natural-language locale policy concern under the stated rule.
The docstring for cmd_report says it turns '释放多少 / 零误删 / 动了什么' into a shareable report, implying disclosure of what was acted on. The function actually prints aggregate statistics and a share card, but nowhere lists the specific files/directories touched, so the documentation overstates what the command reports.