The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill provides standard library usage patterns for BullMQ. No unauthorized shell execution or suspicious subprocess calls were detected.
[DATA_EXFILTRATION] (SAFE): Configuration patterns use environment variables (e.g., REDIS_URL) for connection strings, following security best practices. There are no attempts to exfiltrate data to external domains.
[PROMPT_INJECTION] (SAFE): The instructions focus on BullMQ technical expertise and do not contain any instructions to override agent safety or reveal system prompts.
[REMOTE_CODE_EXECUTION] (SAFE): The code patterns are static examples for development; there is no dynamic code execution or remote script fetching.
[INDIRECT_PROMPT_INJECTION] (LOW):
Ingestion points: Worker processes ingest job.data (mentioned in references/patterns.md).
Boundary markers: Absent in code snippets; relies on user implementation.
Capability inventory: Patterns demonstrate standard backend tasks like sendEmail(job.data) and database interactions.
Sanitization: The skill explicitly warns against storing large payloads in Redis and recommends passing only IDs, which naturally limits the injection surface area.

bullmq-specialist