The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a protocol for processing external data (user input), which constitutes a potential surface for indirect prompt injection findings.
Ingestion points: The skill ingests the user's prompt and conversation history as the 'information field' to be scanned (SKILL.md).
Boundary markers: No specific delimiters or 'ignore' instructions are provided to isolate untrusted content during the observation phase.
Capability inventory: The skill's capabilities are restricted to internal cognitive processing and text response generation; no file system access, network operations, or shell tools are utilized.
Sanitization: There is no evidence of input validation or content sanitization processes.
[SAFE]: The skill contains no executable scripts, binary files, or external package dependencies.
[SAFE]: External URLs found in the metadata and documentation point to the official infrastructure of the known developer (InfraNodus), posing no security risk.

vipassana-llm