The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill framework is designed to ingest and analyze datasets provided by external users, which presents a surface for indirect prompt injection.
Ingestion points: The workflow in SKILL.md and examples/sample_scenario.md involves processing user-supplied CSV or tabular data.
Boundary markers: There are no explicit boundary delimiters defined to separate raw data values from the agent's internal instruction context.
Capability inventory: The skill utilizes standard statistical and data manipulation capabilities via pandas, numpy, and scikit-learn as described in references/missing-data.md.
Sanitization: The instructions do not specify a requirement to sanitize or filter data content before the agent performs analysis.

stat-eda