The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data and configurations to generate visualization files, creating a surface for injection attacks.
Ingestion points: The generate_charts_from_data function and individual chart functions in SKILL.md accept data structures and chart_configs objects.
Boundary markers: Absent. There are no delimiters or instructions to ignore potential commands embedded in the processed data.
Capability inventory: The skill uses plt.savefig (Matplotlib), fig.write_html (Plotly), and os.makedirs (Python os module) to create directories and write files to the local filesystem.
Sanitization: Absent. The skill does not validate or sanitize the output path in chart_configs, which could allow an attacker to overwrite sensitive files via path traversal (e.g., using ../../.bashrc as a filename). Additionally, interactive HTML charts generated by Plotly are not explicitly sanitized, creating a potential Cross-Site Scripting (XSS) surface if the input data contains malicious scripts.

chart-generator