The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill handles user prompts and documents for building RAG pipelines, creating a risk surface where untrusted input could influence agent behavior.
Ingestion points: User-supplied <prompt> and external text documents indexed via mcp__plugin_ruflo-core_ruflo__ruvllm_hnsw_add in SKILL.md.
Boundary markers: Absent; there are no instructions or delimiters specified to isolate untrusted text or prevent the execution of embedded instructions.
Capability inventory: Relies on specialized prompt formatting and vector search tools, along with access to the Bash tool.
Sanitization: Absent; no input sanitization, filtering, or validation controls are defined within the skill instructions.

chat-format