skills/maragudk/skills/marimo/Gen Agent Trust Hub

marimo

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICAL
Full Analysis
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill provides instructions for installing the marimo library and its optional dependencies via pip. marimo is a well-known open-source project and the installation instructions provided are standard and benign.\n- [DYNAMIC_EXECUTION] (LOW): The skill documents the use of mo.persistent_cache, which utilizes Python's pickle module for disk-based serialization. While pickle can be unsafe if used with untrusted data, this is an expected feature for the documented environment's primary purpose of local data science development.\n- [INDIRECT_PROMPT_INJECTION] (LOW): The skill documents surfaces for ingesting external data (e.g., CSV, SQL, remote URLs) into a notebook context. 1. Ingestion points: Documented in references/sql.md (remote CSV/S3) and references/media.md (remote images/audio). 2. Boundary markers: Not explicitly defined in basic code examples, but security isolation via mo.iframe is documented in references/html.md. 3. Capability inventory: The tool possesses capabilities for network operations (aiohttp), file writing (open), and command execution (CLI tools), which are standard for a development notebook. 4. Sanitization: references/html.md provides explicit guidance on using html.escape() for user-provided content to prevent XSS.\n- [SAFE] (SAFE): The scanner alert for a blacklisted URL in sql.md refers to https://example.com/data.csv, which is a standard documentation placeholder and constitutes a false positive.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 05:47 PM