The Agent Skills Directory

[SAFE]: The skill consists of behavioral instructions that guide the AI agent to be more efficient with tokens without compromising technical accuracy.
[COMMAND_EXECUTION]: The instructions encourage the agent to suggest standard command-line tools (such as git, gh, aws, prettier, and eslint) to the user as alternatives to expensive LLM processing for routine tasks.
[EXTERNAL_DOWNLOADS]: The README provides links to several third-party GitHub repositories for cost-monitoring utilities. These are presented as informational resources for the user and are not automatically installed or executed by the skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to activate when the user discusses topics like budget or spending. This is a standard functional trigger and does not pose a security risk.

cost-mode