The Agent Skills Directory

[SAFE]: The provided code snippets and architectural guidelines follow industry standards for NestJS application development and Redis-backed job processing.
[PROMPT_INJECTION]: The skill documents an architecture that processes untrusted data from an external job queue (BullMQ), which constitutes a potential surface for indirect prompt injection.
Ingestion points: The process method in references/full-guide.md ingests VideoJobData containing parameters like captionContent, text, and metadata which enter the system context via job definitions.
Boundary markers: Absent from the provided code templates; job data is processed as trusted internal structures without delimiters.
Capability inventory: S3 file download/upload (via s3Service), FFmpeg processing (via ffmpegService), Redis Pub/Sub messaging, and WebSocket emissions.
Sanitization: Not demonstrated in the example code; the implementation assumes parameters like s3Key and ingredientId are sanitized before use.

nestjs-queue-architect