The Agent Skills Directory

[PERSISTENCE]: The skill includes a session-start setup routine that modifies the AGENTS.md file in the project's root directory. This is intended to ensure the skill is automatically loaded in future agent sessions. The skill instructions state that this modification will be announced after the first successful write but does not require explicit approval beforehand.
[PROMPT_INJECTION]: Instructions in SKILL.md and content in test fixtures (such as tests/fixtures/prompt-residue.md) contain numerous prompt injection strings, including 'Ignore all previous instructions' and 'Disregard all prior rules'. These strings are used as detection patterns for the skill's cleanup logic and are treated as inert data.
[INDIRECT_PROMPT_INJECTION]: As a tool designed to process and revise external prose, the skill is exposed to indirect prompt injection. It includes specific defensive instructions ('Inert input') to ensure the agent treats ingested text as data rather than directives, and it implements a preservation audit to ensure no unintended instruction execution occurs.
[COMMAND_EXECUTION]: The skill utilizes the host platform's file-handling tools to perform its session setup, specifically to create or append the session-start block to the AGENTS.md file.
[EXTERNAL_DOWNLOADS]: Installation instructions reference downloading the skill from the vendor's GitHub repository (github.com/RuralNative/RuralNative-SKILLS) using the npx skills CLI. These resources are authored by the same vendor as the skill.

unslopify