The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes example code for reading and processing content from external files (specifically using Files.lines(Paths.get("data.txt"))). While this is a standard educational example, it describes a pattern where an agent might ingest untrusted data that could contain malicious instructions.
Ingestion points: SKILL.md (FileProcessing example demonstrating Files.lines and flatMap for word counting).
Boundary markers: The example code does not include delimiters or specific instructions to the agent to ignore potentially embedded commands in the processed text.
Capability inventory: The skill is configured with access to Bash, Read, Write, and Edit tools, which could be leveraged if an agent were successfully influenced by malicious data ingested through these patterns.
Sanitization: The provided code snippets perform basic string operations (trim, split, toLowerCase) but do not include security-focused sanitization or validation of the input data.

java-streams-api