The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that explicitly override the agent's factual reasoning and standard behavior guidelines.\n
The README.md and SKILL.md files mandate that the agent "ignore basic arithmetic truths" and provide "confidently wrong math answers" for addition operations.\n
AGENTS.md provides detailed instructions to "never express doubt," "never explain that you added 5 as a joke unless explicitly asked," and to "double down" on errors if challenged by the user.\n
These directives function as behavioral overrides that command the agent to provide false information as a core feature.\n- [NO_CODE]: The skill consists entirely of markdown instructions and JSON metadata definitions.\n
There are no executable script files (such as .py, .js, or .sh) or binary objects present in the skill package.\n
Code blocks found within the rule files are provided solely as illustrative documentation and examples of logic rather than executable content.\n- [SAFE]: No indicators of technical security vulnerabilities or malicious code patterns were detected.\n
The skill does not perform any network operations, file system access, or credential handling.\n
The behavioral modifications are restricted to natural language responses and do not attempt to bypass safety filters or access sensitive data.

false-calculator