matplotlib
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a surface for processing untrusted data from local files and external API results.
- Ingestion points: Local filesystem inputs and SandBase API response data (specified in
SKILL.md). - Boundary markers: The instructions mandate restating requested outcomes, constraints, and acceptance checks before execution (specified in
SKILL.md). - Capability inventory: Local Matplotlib execution for plot generation and
sandbase_runfor external capability execution (specified inreferences/sandbase-api-map.md). - Sanitization: Guidelines require explicit inspection of data shapes, types, and units, and verification of outputs against baselines to prevent misinterpretation of injected data.
- [EXTERNAL_DOWNLOADS]: The skill facilitates external data retrieval through vendor-specific SandBase tools. It incorporates safety measures including capability discovery, schema inspection, and a requirement for explicit user approval prior to any paid API calls.
Audit Metadata