streamlit-snowflake
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill includes various CLI commands for the Snowflake CLI (
snow), such assnow streamlit deployandsnow stage put, as well asopensslcommands for generating RSA keys. These are standard developer tools for the intended workflow. - [EXTERNAL_DOWNLOADS]: The templates (e.g.,
templates-container-runtime/requirements.txtandtemplates/environment.yml) reference downloading Python packages from the Snowflake Anaconda Channel and PyPI. These are well-known, trusted registries for software dependencies. - [INDIRECT_PROMPT_INJECTION]: The provided templates, such as
data_explorer.pyandstreamlit_app.py, ingest user input via Streamlit widgets to execute SQL queries. This constitutes a potential attack surface; however, the skill explicitly provides mitigation logic including regex-based identifier validation and identifier quoting to prevent SQL injection. - [CREDENTIALS_UNSAFE]: The
references/authentication.mdfile provides instructions on managing Snowflake credentials and mentions sensitive paths like~/.snowflake/connections.toml. The documentation follows security best practices by advising against hardcoding passwords and recommending the use of environment variables or key-pair authentication.
Audit Metadata