The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes npx prismic, which downloads and executes the latest version of the Prismic CLI from the NPM registry. This is the official tool provided by the vendor for repository management and is a standard development practice.\n- [COMMAND_EXECUTION]: The agent is instructed to run npx prismic commands autonomously, specifically for creating new repositories, without seeking user confirmation. This increases the risk of unintended resource creation or operational errors.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes functionality to read external documentation (npx prismic docs view), creating a vulnerability surface where instructions embedded in retrieved content could influence agent behavior.\n
Ingestion points: Documentation content read via the docs view <path> command.\n
Boundary markers: The instructions do not define delimiters or provide warnings for the agent to distinguish documentation text from its core instructions.\n
Capability inventory: The agent has access to the Bash tool to execute npx prismic commands, which can manage content models and repository settings.\n
Sanitization: No sanitization or validation of the retrieved documentation content is specified in the skill.\n- [COMMAND_EXECUTION]: The skill requires passing the user's natural language intent as a command-line argument (--user-intent "<the user's request>") to every CLI call. This pattern creates a potential risk of shell command injection if the user's input contains characters that can break out of the double-quoted string (such as backticks or nested quotes) and the platform does not provide automatic shell escaping.

prismic