data-viz-renderer
Warn
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/build_infographic.pyfacilitates arbitrary file writes to paths specified in the JSON configuration. - Evidence: The script extracts the
outputfield from the user-provided JSON (config.get("output", "infographic.html")) and uses it directly inopen(output_path, "w")without sanitization or directory sandboxing. - Risk: An attacker could use path traversal (e.g.,
../../.ssh/authorized_keys) to overwrite sensitive configuration or system files with the generated HTML content. - [INDIRECT_PROMPT_INJECTION]: The skill processes structured JSON data from potentially untrusted sources to generate HTML infographics.
- Ingestion points: The main script
scripts/build_infographic.pyreads data from files orstdin. - Boundary markers: Absent. The script assumes the provided JSON is trustworthy data.
- Capability inventory: The script has the capability to write files to the disk (
open,f.write). - Sanitization: The script correctly uses
html.escape(aliased as_esc) to sanitize all text fields (titles, labels, values) before interpolating them into the HTML template, effectively mitigating Cross-Site Scripting (XSS) risks.
Audit Metadata