The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data files (WPILOG), creating a potential surface for indirect prompt injection if the logs were to contain malicious instructions designed to be interpreted by the AI.
Ingestion points: Local binary log files are read into the environment using node:fs/promises for processing by the wpilog-parser library.
Boundary markers: No specific delimiters or instructions to ignore embedded content are used when reading the log files.
Capability inventory: The skill's documented capabilities are limited to reading local files and logging output to the console; there are no network operations, file-write capabilities, or arbitrary execution paths that would allow an injection to perform dangerous actions.
Sanitization: Content is parsed according to the WPILOG specification by the library, but no specific sanitization for LLM instructions is applied to the decoded data.

analyze-wpilog