The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of standard shell utilities (such as wc, grep, awk, find, and cat) to analyze file sizes, extract section headers, and manage the file system during the optimization process. These commands are used as intended for developer workflows.
[DYNAMIC_EXECUTION]: The skill encourages the conversion of embedded markdown code blocks into standalone executable scripts within a scripts/ directory. This is a recommended architectural pattern for the platform to reduce the token count in the primary instruction set while maintaining functionality.
[INDIRECT_PROMPT_INJECTION]: As a tool designed to process and optimize other skill files, it technically has an ingestion surface for untrusted local data. However, the logic is focused on restructuring and metadata enrichment rather than automated execution of the content it processes.
[SAFE]: While documentation examples in REFERENCE.md reference sensitive environment variables like DATABASE_URL, REDIS_URL, and API_KEY, these appear strictly as placeholders in templates for validation and diagnostic scripts. There are no patterns suggesting the exfiltration of these values or hardcoding of actual secrets.

skill-optimizer