Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent Chinese-language vocational lesson-plan evaluation helper with optional document export, and I did not find hidden persistence, data theft, or deceptive behavior.
Install this if you want Chinese-language vocational lesson-plan critique and scoring. Before using the .docx export path, be aware it may require Pandoc and may create a local Word file; review any package-install command before running it.
The instruction to activate whenever the user mentions lesson-plan quality creates an ambiguous trigger boundary that is too broad for safe routing. In multi-skill environments, this can lead to persistent false activations, capture of unrelated pedagogical discussions, and unintended application of the skill's workflow and output requirements where the user did not request them.
The instruction to activate whenever the user mentions lesson-plan quality creates an ambiguous trigger boundary that is too broad for safe routing. In multi-skill environments, this can lead to persistent false activations, capture of unrelated pedagogical discussions, and unintended application of the skill's workflow and output requirements where the user did not request them.
Line L160 says not to load references/scoring-standard.md unless the user asks for detailed scoring-calculation details. But L164-L166 defines a scoring mechanism for the standard report and points to that file as the source of the full dimension scores and deduction rules, which conflicts with the earlier instruction and creates intent ambiguity about whether the file should be consulted during normal report generation.
The skill explicitly instructs the agent to run a shell command (pandoc ...) to create a local .docx file, even though the core skill purpose is content evaluation rather than system-level file operations. This expands the skill from analysis into command execution and file creation, which can create side effects, violate least privilege, and become dangerous if filenames, paths, or surrounding command construction later incorporate untrusted input.
The skill directs generation of a .docx via Pandoc without disclosing that this creates local files and may invoke command execution side effects. Even if the command is fixed in the documentation, the absence of a user-facing safety notice and execution boundary makes the behavior riskier, especially in systems where agents may act on operational instructions automatically.
The title and opening description define the report output format entirely in Chinese, implying the skill's outputs must be produced in that language. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless a justified locale constraint is documented, which is not present here.
Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.
# Ubuntu/Debian
sudo apt install pandoc
# macOS
brew install pandoc
This markdown file is entirely written in Chinese and does not indicate that the language is optional, selectable, or limited to a justified Chinese-only context. Under the policy rule for natural-language violations, forcing a specific language without user opt-in can be a locale-policy issue.
No suspicious patterns detected.