The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted CSV data and generates HTML output without adequate sanitization, creating a Cross-Site Scripting (XSS) vulnerability surface.
Ingestion points: scripts/main.py reads data from CSV files provided as CLI arguments.
Boundary markers: The skill does not use delimiters or instructions to prevent the agent or the generated output from interpreting data as code.
Capability inventory: The script performs file write operations to the local filesystem using Path(output).write_text(html).
Sanitization: The generate_dashboard_html function in scripts/main.py interpolates metric names (extracted from CSV headers) directly into HTML strings without escaping, which could execute malicious scripts if the generated dashboard is opened in a browser.
[METADATA_POISONING]: The SKILL.md file contains a misleading statement in the 'Skill Boundaries' section, claiming the skill 'Cannot access your actual data.' This is inaccurate, as the primary functionality of the script is to read and process local CSV data files provided by the user.

data-visualizer