The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends using npx @hacksmith/doraval as a validation gate. This executes code from a third-party npm package that is not part of the established trusted vendor list.
[COMMAND_EXECUTION]: The skill instructions involve executing several CLI tools, including dora and claude, to perform plugin validation, testing, and hot-reloading within the local development environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and restructure user-provided plugin directories, manifest files, and skill definitions, which creates an entry point for indirect prompt injection.
Ingestion points: Project files such as plugin.json and SKILL.md are read during the 'Structure' and 'Package' phases.
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to the agent to treat user-provided plugin content as untrusted data.
Capability inventory: The skill possesses the capability to execute shell commands (dora, npx, claude) and navigate the file system.
Sanitization: There are no explicit steps defined to sanitize or validate the content of the analyzed plugins before the agent performs operations based on them.

plugin-craft