The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local command-line tools to manage feature metadata.
It specifically uses npm run for commands such as feature:info, feature:validate, and feature:health.
These operations are standard for development workflows and are scoped to the local project environment.
[PROMPT_INJECTION]: The skill is subject to a surface-level risk of indirect prompt injection as it processes external YAML data.
Ingestion points: The agent reads feature manifest files (.yaml) using commands like npm run feature:info.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore content within the manifests.
Capability inventory: The skill possesses Bash, Write, and Edit permissions, which are necessary for its documented functionality.
Sanitization: There is no mention of sanitizing or validating the text content within the manifest files before processing.

feature-manifest