The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external files that could contain hidden or malicious instructions meant to influence the agent's behavior.
Ingestion points: The agent reads user-provided targets such as .tex files, PDFs, course notes, and project metadata including reviews/INDEX.md and CLAUDE.md in Phase 0.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' commands to isolate ingested data from the system prompt.
Capability inventory: The skill utilizes Bash(cat*) and Bash(grep*) to read files, Write to create study sheets, and AskUserQuestion for user interaction.
Sanitization: There are no instructions provided to sanitize, filter, or validate the content extracted from the external files before processing.

grill-me