The Agent Skills Directory

[SAFE]: The skill implements proactive security measures for the BEUI_PRO_TOKEN, including instructions to never print the token, commit it to source files, or accept it through user prompts. Access is managed via environment variables.\n- [EXTERNAL_DOWNLOADS]: The skill fetches registry metadata from https://pro.beui.dev to discover available components. This is a legitimate vendor resource necessary for the skill's operation.\n- [REMOTE_CODE_EXECUTION]: The skill uses the shadcn CLI to install components directly from the vendor's authenticated registry. While this involves downloading and executing external source code, it is an expected behavior for this type of development tool.\n- [COMMAND_EXECUTION]: Shell commands such as curl are used for registry lookups, and npx, pnpm dlx, or bunx are utilized to run the shadcn utility for component management.\n- [PROMPT_INJECTION]: The skill includes a surface for indirect prompt injection by processing external JSON data from https://pro.beui.dev/r/registry.json.\n
Ingestion points: Data enters the context via curl requests to the remote registry.\n
Boundary markers: No specific markers are defined to isolate registry data from instructions.\n
Capability inventory: The agent can execute commands and install packages via CLI tools.\n
Sanitization: The agent is instructed to use only specific slugs derived from the registry, which limits the potential for executing arbitrary payloads.

beui-pro