The Agent Skills Directory

[COMMAND_EXECUTION]: The instructions direct the agent to execute the canvaslms CLI tool using shell commands. The usage of user-influenced strings, specifically regex filters (e.g., canvaslms courses "regex") and selectors (e.g., -a "assignment regex"), creates a risk of shell command injection if the agent or the underlying execution environment does not correctly escape these parameters before shell interpolation.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and display content from external Canvas LMS sources, such as announcements, wiki pages, and discussion threads. This content is authored by third parties (instructors or other students) and may contain hidden instructions intended to hijack the agent's behavior during the synthesis phase.
Ingestion points: Data is retrieved through canvaslms modules view, canvaslms pages view, canvaslms assignments view, and canvaslms discussions view in SKILL.md.
Boundary markers: The instructions do not specify any delimiters or safety warnings for the agent to treat the retrieved content as untrusted data.
Capability inventory: The skill possesses shell execution capabilities for the canvaslms tool and potentially other system commands if injection occurs.
Sanitization: There are no instructions for sanitizing or escaping the HTML/Markdown content fetched from Canvas before processing.

canvas-info