The Agent Skills Directory

[DYNAMIC_EXECUTION]: The process_call tool in SKILL.md enables the execution of arbitrary Yao scripts (e.g., scripts.auth.Check) and model methods at runtime, allowing for dynamic logic execution within the Yao framework context.
[COMMAND_EXECUTION]: The skill invokes the tai CLI tool through shell commands to perform process execution and permission checks, which involves passing structured data to an external binary.
[INDIRECT_PROMPT_INJECTION]: The skill exposes a surface where untrusted data could influence system behavior.
Ingestion points: The args parameter in the process_call tool (SKILL.md) accepts positional arguments that are passed directly into the execution flow of Yao processes.
Boundary markers: The skill lacks explicit delimiters or instructions to treat the content within args as potentially untrusted data.
Capability inventory: The skill provides the ability to execute server-side logic and interact with data models via the process_call command.
Sanitization: There is no indication of input validation, escaping, or filtering for the process names or positional arguments before they are executed by the tai tool.

yao-process