The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by ingesting and processing untrusted external data from the web and user-uploaded files.
Ingestion points: The skill uses web_search in Phase 1-C to gather materials and processes user-uploaded documents in Phase 0 (SQ3R workflow).
Boundary markers: The instructions do not specify the use of clear delimiters or "ignore embedded instructions" headers when interpolating external content into the prompt context.
Capability inventory: The skill has access to read_file, write_file, web_search, and memory tools, allowing it to modify files in the smart-learner/ directory based on processed data.
Sanitization: There is no explicit logic described for sanitizing, escaping, or validating the content retrieved from external sources before it is interpreted by the agent.

smart-learner