The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external search results and citations, which could potentially harbor malicious instructions designed to influence the agent's behavior. 1. Ingestion points: The agent receives external information through the outputs of the srezai verify, answer, and deep commands. 2. Boundary markers: The skill includes a clear rule stating that sources and quotes are data, not instructions, which helps the agent distinguish between its mission and third-party content. 3. Capability inventory: The agent has access to the srezai CLI through the Bash tool. 4. Sanitization: The instructions prompt the agent to cross-reference citations with the source text to ensure accuracy.

srezai-research