claude-certification
Fail
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill defines a workflow to fetch Python scripts from a remote repository and execute them on the host system. Evidence: Instructions to fetch files from github.com/fancyboi999/ai-engineering-from-scratch-zh and execute via python3 /code/main.py.
- [COMMAND_EXECUTION]: The skill instructs the agent to run system commands and unit tests. Evidence: python3 -m unittest discover -s /code/tests -v.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests remote JSON and Markdown content into the agent context without sanitization. Ingestion points: certifications/claude/program.json, tracks/*.json, docs/zh.md, quiz.json. Boundary markers: Absent. Capability inventory: Subprocess execution, local file system access. Sanitization: Absent.
- [EXTERNAL_DOWNLOADS]: Fetches configuration and code artifacts from third-party URLs including github.com and aieng-zh.cn.
Recommendations
- AI detected serious security threats
Audit Metadata