The Agent Skills Directory

[SAFE]: The skill provides standard documentation and Python code snippets for natural language processing using the 'spaCy' library. No malicious activities like unauthorized network calls or credential harvesting were detected.- [SAFE]: Code examples for model training and inference represent standard industry usage patterns and do not request excessive system privileges.- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection in the advanced LLM-based extraction examples. Ingestion points: 'text' parameter in 'llm_extract_entities' and 'extract_relations' functions in SKILL.md. Boundary markers: The functions use double quotes ("{text}") which offer minimal protection against adversarial input designed to override instructions. Capability inventory: The skill performs extraction and relation identification via LLM calls. Sanitization: No input validation or escaping is present in the examples to mitigate prompt manipulation risks.

Entity Extractor