The Agent Skills Directory

[SAFE]: The skill implements a comprehensive analytical framework using educational reference documents to guide the agent through game theory concepts such as Nash Equilibrium and archetypal game structures.\n- [SAFE]: A thorough review of the skill instructions and reference files found no instances of prompt injection, obfuscation, or unauthorized access to sensitive data.\n- [DATA_EXFILTRATION]: The skill is configured to use a web search tool to gather contextual information for its analysis. This is a legitimate functional feature for data enrichment and does not target sensitive user information.\n- [COMMAND_EXECUTION]: The skill writes analysis results to markdown files within a local directory. This is a standard reporting function and does not involve the execution of arbitrary system commands or scripts.\n- [PROMPT_INJECTION]: The skill processes external data retrieved from web searches, creating a surface for indirect prompt injection. However, the risk is categorized as safe/low because the agent's capabilities are restricted to generating static markdown reports. Evidence chain: Ingestion points include web search results; Boundary markers are not explicitly defined in the prompt instructions; Capability inventory is limited to local markdown file writing; Sanitization procedures are not specified.

game-theory-lens