The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as an ingestion point for untrusted data from web search results and project files, creating a potential surface for indirect prompt injection attacks.
Ingestion points: The orchestration logic involves calling zai-web-search and zread.read to gather context.
Boundary markers: The instructions do not define clear delimiters or specify that the agent should ignore instructions embedded within the data retrieved from tools.
Capability inventory: The skill has the ability to invoke vision tools, perform web searches, and analyze file structures.
Sanitization: The logic does not include mechanisms to validate or sanitize external content before it is incorporated into the agent's context.
[EXTERNAL_DOWNLOADS]: The installation guide directs users to download the skill and its dependencies from the author's GitHub repository (tianxiao1430-jpg/zai-skills).

zai-orchestrator