The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to interact with a local secrets CLI tool using the Bash tool. It provides specific command templates (e.g., secrets add <service>, secrets status) to ensure consistent and safe interaction with the underlying credential vault.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided input through the $request argument and incorporates this data into shell commands executed in the local environment.
Ingestion points: The $request argument defined in the frontmatter of SKILL.md is the primary entry point for untrusted data.
Boundary markers: The skill does not employ specific delimiters or "ignore instructions" markers when interpolating user data into commands.
Capability inventory: The skill utilizes the Bash tool to execute arbitrary shell commands and the Read tool to inspect local file content.
Sanitization: While the instructions direct the agent to "Resolve the requested action" and use specific command shapes, there is no explicit requirement for shell-escaping or validation of the service names provided by the user before they are passed to the CLI.

secrets