zhongkao-english
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions and the 'publish-quiz.md' script direct the agent to execute shell commands such as 'git add', 'git commit', 'git push', and 'gh api'. These commands are used to automate the deployment of generated HTML quiz files to the author's GitHub repository and configure GitHub Pages.
- [EXTERNAL_DOWNLOADS]: The generated HTML output includes a reference to an external JavaScript file ('feishu-sync.js') hosted on the author's GitHub Pages domain. This script is loaded into the user's browser when they open the generated quiz file.
- [DATA_EXFILTRATION]: The skill includes functionality via the 'feishu-sync.js' script to transmit quiz results, including scores, accuracy, and practice metadata, to an external service (Feishu/Lark). This is presented as a telemetry and result-tracking feature for students and teachers.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (topics, grades, question counts) to generate quizzes.
- Ingestion points: User-specified parameters are identified in the 'Workflow' section of 'SKILL.md'.
- Boundary markers: Explicit delimiters for separating user-supplied topics from internal generation instructions are not strictly defined.
- Capability inventory: The skill has the ability to execute shell commands ('git') and write files to the local file system.
- Sanitization: 'SKILL.md' (Step 4) and 'publish-quiz.md' (Step 2) contain instructions for the agent to sanitize input by replacing spaces and removing special characters before incorporating them into file names or shell command arguments, which mitigates the risk of command injection.
Audit Metadata