mcpa-certification
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill downloads Python scripts from the author's GitHub repository and executes them locally using
python3to facilitate interactive labs and automated validation. - [EXTERNAL_DOWNLOADS]: Fetches curriculum data, JSON configuration tracks, and validation scripts from the
fancyboi999/ai-engineering-from-scratch-zhrepository on GitHub. - [COMMAND_EXECUTION]: Invokes shell commands to execute Python scripts (
python3 <lesson-path>/code/main.py) and run unit tests (python3 -m unittest discover) as part of the lesson workflow. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted lesson content and metadata from an external source which influences the agent's behavior during mentoring sessions.
- Ingestion points: Reads lesson docs (
zh.md), JSON tracks, and scripts from a remote GitHub repository. - Boundary markers: The instructions mandate strict adherence to the track JSON and provide specific logic for quizzes and recollection steps to constrain the session.
- Capability inventory: Execution of Python scripts via
python3and unittest across various lesson paths. - Sanitization: None specified for the content retrieved from the remote repository.
Audit Metadata