The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill accepts external data which could potentially harbor malicious instructions.
Ingestion points: User-supplied data is ingested via the --input and --input-file flags of the tilebox job submit command in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard embedded commands within the input data.
Capability inventory: The skill can execute various shell commands through the tilebox CLI, including job submission, retries, and cancellation.
Sanitization: The instructions do not describe any sanitization or validation of the input data before it is processed by the CLI.
[DYNAMIC_EXECUTION]: The skill documentation includes an example of dynamic Python execution using a shell heredoc (`python
<<'PY'`) to facilitate task serialization.
[COMMAND_EXECUTION]: The skill relies on the execution of host-level shell commands via the tilebox CLI to perform job management tasks.

tilebox-workflow-jobs