The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute specific shell commands including cat, curl, and find. These are used to inspect internal registry files (/data/previews.json), verify local port availability on localhost for diagnostic purposes, and scan the workspace to identify project structures.
[INDIRECT_PROMPT_INJECTION]: The diagnostic workflow involves reading and processing data from internal registry files and project-specific files (such as package.json or index.html). This content is used to populate parameters for the preview_serve tool. While this creates an ingestion surface for potentially untrusted workspace data, it is a functional requirement for the skill's primary purpose of managing local development previews.
Ingestion points: Files at /data/previews.json, /data/preview_history.json, and project metadata files within /data/workspace.
Boundary markers: None explicitly defined in the instructions for the registry data.
Capability inventory: Uses cat and find for file access, and preview_serve for executing build/start commands.
Sanitization: The skill assumes the validity of commands found in the registry or history, relying on the underlying platform's tool safety for execution.

browser-preview