The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the installation of external code and components through commands like npx shadcn@latest add https://ui.inference.sh/r/chat.json and npx skills add for several remote repositories (belt-sh/cli, inference-sh/skills@agent-ui). These patterns integrate external, third-party code into the project environment.- [EXTERNAL_DOWNLOADS]: Fetches component configurations and UI blocks from the ui.inference.sh domain and refers to documentation on inference.sh.- [PROMPT_INJECTION]: The skill implements a chat interface that processes user input and assistant messages, which serves as a vector for indirect prompt injection if external data is not sanitized.
Ingestion points: Data enters the system via the ChatInput component and is rendered in the ChatMessage component in SKILL.md.
Boundary markers: The provided usage examples do not include delimiters or instructions to the agent to disregard instructions within the message content.
Capability inventory: The skill provides UI components and does not include scripts with high-privilege system access or dangerous file operations.
Sanitization: The implementation snippets do not show explicit sanitization or escaping of the message content before rendering.

chat-ui