teaching-pipeline
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes several local Python utilities (check_passport.py, check_alignment_gate.py, check_quality_gate.py, check_content_markers.py, and build_dashboard.py) to perform deterministic validation of course data and generate visual dashboards. These scripts are executed via the shell using python3.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted data from student evaluations and external feedback to inform the course improvement cycle. \n- Ingestion points: Evidence is gathered from evaluation analysis themes and student feedback batches in Stage 5 and Stage 6. \n- Boundary markers: The skill does not explicitly define delimiters or specific 'ignore' instructions for the LLM when processing this external text. \n- Capability inventory: The agent has access to file writing (course_passport.yaml) and command execution (local scripts). \n- Sanitization: Although the skill implements PII guards and triangulation, it lacks technical sanitization for the prompt interpolation of external student-provided content.
Audit Metadata