alterlab-seaborn
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external datasets (e.g., CSV files, Pandas DataFrames) which serves as an ingestion point for untrusted data into the agent's context.
- Ingestion points: Data input occurs when the agent uses Pandas to read user files (e.g.,
pd.read_csv('data.csv')mentioned inexamples.md) or process existing DataFrames for plotting. - Boundary markers: The instructions do not explicitly provide delimiters or specific warnings for the agent to ignore potentially malicious instructions embedded within the data (e.g., in column headers or string values).
- Capability inventory: The skill uses
Bash(python:*)to execute Python code, which includes library imports and file system access (plt.savefig,pd.read_csv). - Sanitization: There is no mention of sanitizing or validating the content of the DataFrames before they are processed by plotting functions or interpolated into descriptions.
Audit Metadata