The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes data from external census queries, which constitutes a potential surface for indirect prompt injection.
Ingestion points: The skill is designed to receive results from ACS census queries for evaluation and classification (specified in SKILL.md).
Boundary markers: The instructions lack explicit delimiters or guardrails to ensure the agent distinguishes between statistical data and potential instructions embedded within the query results.
Capability inventory: The skill is granted access to high-privilege tools, specifically 'Bash' and 'Read', which could be targeted if malicious strings were present in the ingested data.
Sanitization: No input validation or sanitization logic is provided in the skill files to filter or escape content from the external census data before processing.

interpreting-quality