The Agent Skills Directory

Prompt Injection (SAFE): No instruction override or bypass patterns detected.
Data Exposure (SAFE): No hardcoded credentials or sensitive file access patterns found. The templates use external configuration for connections.
Remote Code Execution (SAFE): While the skill instructions include npx tsx for running workers, this is a standard development practice. No piped remote execution or untrusted source downloads are present.
Indirect Prompt Injection (LOW): The worker template processes job.data which is an untrusted ingestion point.
Ingestion points: job.data in templates/worker.ts.
Boundary markers: None.
Capability inventory: No unsafe tool calls are performed in the provided template.
Sanitization: The template only performs console.log on the data. Users should be advised to sanitize data if the processor performs shell commands or database writes.

worker-patterns