iblai-api-canvas-course-builder
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it is designed to ingest and process external course specification files (YAML or JSON) to automate course creation.
- Ingestion points: The
scripts/build_course.pyscript reads a user-provided specification file via theload_specfunction. - Boundary markers: There are no explicit boundary markers or instructions within the processing logic to ignore embedded instructions in the specification file content.
- Capability inventory: The skill possesses significant capabilities, including performing network operations (POST, PUT, GET, DELETE) against the Canvas REST API and performing file uploads as implemented in
scripts/canvas_client.py. - Sanitization: The skill relies on server-side sanitization performed by the Canvas LMS API itself for HTML content (e.g., in pages and assignments), as noted in the
references/content-types.mddocumentation. - [SAFE]: The skill demonstrates excellent security posture by explicitly instructing users not to hardcode API tokens or include them in output. The code retrieves sensitive credentials (CANVAS_API_TOKEN) from the environment, which is a standard security best practice. Furthermore, the implementation of the Canvas API client includes handling for Canvas-specific security measures, such as the
while(1);JSON hijacking prevention prefix.
Audit Metadata