The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and installs necessary Python dependencies (ydata-profiling, duckdb, datasketches) from PyPI and references the Chart.js library from a well-known CDN (cdnjs.cloudflare.com) to render visualizations. These are standard operations for data analysis tools and target trusted package registries and services.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied datasets and summarizes them for the AI agent, which is a potential surface for indirect prompt injection. 1. Ingestion points: scripts/analyze.sh, scripts/profile_large.py, and scripts/sketch_ops.py process user-provided CSV, JSON, and Parquet files. 2. Boundary markers: The skill generates structured Markdown and JSON reports to separate data from instructions. 3. Capability inventory: The skill can execute local shell commands and write reports to the filesystem. 4. Sanitization: The summarize_insights.py script filters raw data to provide statistical summaries (means, medians, missing percentages), which acts as a sanitization layer by reducing the risk of the agent interpreting raw text content as instructions.

exploring-data