The Agent Skills Directory

[PROMPT_INJECTION]: The application accepts user-supplied chat messages and forwards them to the OpenAI API, creating a potential surface for indirect prompt injection attacks.
Ingestion points: User content enters the system via the ChatRequest body in ChatController.java.
Boundary markers: The application prepends a system prompt from system-prompt.txt to the message list, but it does not employ explicit delimiters or structural markers (e.g., specific tags or block prefixes) to isolate user input from instructions.
Capability inventory: The skill possesses network capabilities to communicate with the OpenAI API.
Sanitization: There is no evidence of content filtering or sanitization performed on the messages before they are transmitted to the model.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of several packages from the @openuidev scope on npm. These are specific to the OpenUI library described in the skill and are necessary for its functionality.
[COMMAND_EXECUTION]: The setup instructions include running shell commands such as mvn spring-boot:run and npx @openuidev/cli. These are standard development operations for the respective frameworks and use static paths provided in the documentation.

openui-forge-java