The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes a maintenance script scripts/check-updates.py that connects to the official PyPI JSON API (pypi.org) to check for the latest version of the streamlit package. This is a legitimate use of a well-known service for dependency management.
[COMMAND_EXECUTION]: The documentation and installation guides in SKILL.md and references/00-overview.md correctly utilize standard command-line tools for Python environment management, such as pip, uv, and poetry, as well as the framework's own streamlit run command.
[INDIRECT_PROMPT_INJECTION]: The skill documents the creation of LLM-based chat interfaces (references/09-chat-llm.md). This inherently involves ingesting untrusted user input from st.chat_input or st.file_uploader, creating a potential surface for indirect prompt injection in apps built by the user. However, the documentation provides secure patterns for managing these inputs and using environment secrets.
Ingestion points: st.chat_input, st.text_input, and st.file_uploader are documented as primary input methods in references/02-input-widgets.md and references/09-chat-llm.md.
Boundary markers: The educational snippets do not explicitly enforce boundary markers, but they focus on standard framework usage.
Capability inventory: The skill documents capabilities including file writing, network requests, and database access through st.connection in references/11-connections-config.md.
Sanitization: Explicit input sanitization logic is not provided in the reference snippets, as the skill focuses on framework API usage.

streamlit