The Agent Skills Directory

[SAFE]: The skill consists of educational content and best practices for software architecture, specifically regarding data serialization, parser design, and contract enforcement between producers and consumers.
[COMMAND_EXECUTION]: The skill includes a Bash snippet intended for manually verifying file formats. These commands use standard system utilities like grep, wc, shuf, and head to process local text files. This is provided as a diagnostic example for developers and does not involve automated or unsafe execution of untrusted input.
[INDIRECT_PROMPT_INJECTION]: While the skill discusses the parsing of external files, it specifically advocates for defensive programming techniques such as rejecting malformed or empty files to prevent data loss. It does not create a runtime surface for the agent to ingest and execute untrusted data.

import-format-contract-design