The Agent Skills Directory

[SAFE]: The skill contains standard educational content for the Streamlit framework. All demonstrated code snippets follow security best practices, specifically using st.secrets to handle sensitive information like API keys rather than hardcoding them.
[PROMPT_INJECTION]: The skill demonstrates how to handle user-provided text in a chat interface, which represents a potential surface for indirect prompt injection.
Ingestion points: pages/ai_chat.py uses st.chat_input to capture user messages, and pages/submit_feedback.py captures user input via st.text_input.
Boundary markers: The code examples do not include specific delimiters to separate user content from instructions when calling the OpenAI API.
Capability inventory: The snippets focus on UI display and do not include high-risk capabilities like arbitrary command execution or file system modification based on the user input.
Sanitization: No input sanitization or filtering is implemented in the instructional examples.

streamlit