The Agent Skills Directory

[SAFE]: The skill correctly implements and documents background task processing patterns. Analysis found no evidence of prompt injection, data exfiltration, privilege escalation, or persistence mechanisms. The included scripts are benign utilities for environment setup and health monitoring.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external data (e.g., user IDs, email templates, and batch payloads), creating a standard attack surface for indirect prompt injection. As a documentation skill, this is intended behavior and the skill includes recommendations for sanitization.
Ingestion points: Worker processes in SKILL.md and references/bullmq-patterns.md process job.data originating from external sources.
Boundary markers: Not explicitly demonstrated in the code examples.
Capability inventory: File system access (ffmpeg/transcoding), database writes, and network operations (Stripe/Sendgrid APIs).
Sanitization: Referenced as a remediation step in the dead letter queue recovery patterns.
[COMMAND_EXECUTION]: The script scripts/setup_bullmq.sh uses standard npm install commands to fetch well-known libraries (bullmq, ioredis). These operations are essential for the skill's functionality and do not target untrusted or suspicious sources.

background-job-orchestrator