canvas-bulk-grading

Warn

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill defines a workflow (Strategy C) that utilizes the execute_typescript tool to perform bulk grading locally. This allows the agent to generate and execute arbitrary script code to process assignment data, which is a powerful capability that requires careful oversight.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from external student submissions.\n
  • Ingestion points: Student submission data, including file attachments and metadata, are retrieved via the list_submissions tool and processed by the AI.\n
  • Boundary markers: No specific delimiters or instructions are provided to help the AI distinguish between student-provided data and grading instructions.\n
  • Capability inventory: The agent possesses the ability to execute code via execute_typescript and modify student records (grades and comments) via bulk_grade_submissions and grade_with_rubric.\n
  • Sanitization: The skill lacks explicit instructions for sanitizing student-controlled content (such as filenames or submission text) before it is used within the dynamic grading logic or code generation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 3, 2026, 01:19 PM