The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes local plugin files (manifests, agent definitions) provided by the user, which creates a potential surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: Reads configuration and metadata from plugin.json and agent markdown files (agents/*.md).
Boundary markers: Validation rules expect YAML frontmatter to be delimited by --- markers.
Capability inventory: The skill utilizes shell commands (grep, sed, jq) and Python for validation, and includes logic for modifying file permissions (chmod).
Sanitization: Employs yaml.safe_load() to parse frontmatter securely.
[COMMAND_EXECUTION]: The skill provides scripts and instructions to execute various shell commands (find, grep, jq, chmod) for file structure validation and permission management within the local environment.
[DYNAMIC_EXECUTION]: The skill uses Python modules (json.tool and yaml) to dynamically validate the syntax and structure of JSON and YAML files found in the plugin directory.

plugin-validator