The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data into background tasks, creating a potential surface for indirect prompt injection.
Ingestion points: Data enters the system via route handler parameters such as account_id, report_id, and path which are passed to QueueService.enqueue in SKILL.md and references/execution-backends.md.
Boundary markers: The provided examples do not include explicit prompt boundary markers or "ignore instructions" directives when handling these identifiers within the task body.
Capability inventory: The skill enables file system access (e.g., process_import(path)), network dispatch (e.g., CloudTasksExecutionConfig), and database persistence across all configuration examples.
Sanitization: The configuration supports max_argument_identity_bytes in SKILL.md to limit the size of data used for task identity generation, providing resource bounding but no natural language sanitization.
[EXTERNAL_DOWNLOADS]: The skill references the project's source code and documentation on GitHub and provides installation instructions for optional features.
Evidence: Links to the litestar-queues source code, releases, and configuration files on GitHub are present in SKILL.md. Instructions for installing extra dependencies like litestar-queues[cloudrun] and litestar-queues[sqs] are included in references/execution-backends.md.
[COMMAND_EXECUTION]: The skill describes CLI commands for managing worker processes and maintenance tasks.
Evidence: SKILL.md documents commands such as litestar queues run, litestar queues run-maintenance, and litestar queues run-task for claiming and executing records. These are standard administrative tools for the described library.

litestar-queues