The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted community content from Skool and interpolates it into LLM prompts within scripts/skool_rag_query.py. This creates a surface for indirect prompt injection where malicious instructions hidden in community posts could attempt to hijack the agent's response. * Ingestion points: User queries and scraped community content (context) are ingested in scripts/skool_rag_query.py. * Boundary markers: The prompt uses '---' as delimiters but lacks explicit instructions for the LLM to ignore potentially malicious commands within the context block. * Capability inventory: The skill is configured with broad allowed-tools including Bash, Read, Write, and Edit, which could be exploited if an injection is successful. * Sanitization: There is no evidence of sanitization or validation of the retrieved content before it is processed by the LLM.

skool-rag