The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill guides the user to add dependencies from Google Maven (com.google.ai.edge.litertlm) and download pre-trained models from Hugging Face's litert-community. These are well-known and trusted sources in the AI development ecosystem.
[COMMAND_EXECUTION]: Provides instructions for using standard adb commands to push model files to a connected Android device for testing. These commands are standard developer operations and do not represent a security risk when executed in a development context.
[INDIRECT_PROMPT_INJECTION]: The application architecture described in Step 5 and 6 involves a text-based chat interface. User input is passed to an LLM engine, and the output is displayed in the UI. While this represents a potential prompt injection surface for the final application, it is inherent to the skill's purpose of building a chatbot. The skill includes a system instruction ('You are a helpful assistant') as a baseline for the model's behavior.
[REMOTE_CODE_EXECUTION]: There are no patterns of remote code execution. All logic is performed locally on the device using established libraries from trusted vendors.

litert-lm