The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided instructions and schedules them for future execution by passing them verbatim into a prompt for a future agent turn. This creates a surface for instructions to be smuggled into the agent's future context.
Ingestion points: The <instructions> argument parsed in step 1 of SKILL.md.
Boundary markers: The skill uses a textual prefix to identify the scheduled task but does not use formal delimiters or escaping for the user-supplied instructions.
Capability inventory: The skill uses the CronCreate tool to schedule tasks; these tasks can contain any command or tool invocation available to the agent, such as /x-wt-teams as mentioned in the instructions.
Sanitization: Instructions are passed verbatim without filtering, escaping, or validation.
[COMMAND_EXECUTION]: The skill facilitates the future execution of arbitrary instructions through the agent's task harness. It also specifies the use of the date command to compute target execution times.

delay