marimo-notebook
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documents the use of standard package managers and tools to install dependencies. Specifically, it references
uv pip install,playwright install chromiumfor PDF exports, andpip install watchdogfor performance improvements. These are well-known tools from trusted sources used for their intended purpose. - [COMMAND_EXECUTION]: The skill provides instructions for running and testing Python notebooks using the
uv run,marimo, andpytestCLI tools. These commands execute the Python code generated or edited by the agent, which is the primary function of the skill. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to generate Python notebooks based on user-provided logic, creating a surface where malicious instructions could potentially be interpolated into the generated code.
- Ingestion points: User-provided descriptions and logic for notebook cells are processed to generate Python code in
SKILL.mdand referenced templates. - Boundary markers: Not explicitly defined in the templates to separate user logic from the notebook structure.
- Capability inventory: The generated notebooks can perform file system operations, network requests, and execute arbitrary Python logic via the
marimoruntime orpytest(referenced inPYTEST.mdandSKILL.md). - Sanitization: The skill provides structure guidelines but does not define explicit sanitization or escaping mechanisms for user-provided strings within the generated code block.
Audit Metadata