The Agent Skills Directory

[Data Exposure & Exfiltration] (SAFE): No hardcoded credentials or sensitive file paths were detected. The provided code templates use standard Cloudflare Worker environment bindings (c.env.AI, c.env.BUCKET) for secure resource access.
[Unverifiable Dependencies] (SAFE): The templates reference the 'hono' framework, a reputable and standard library for building Cloudflare Workers. No suspicious remote scripts or piped execution patterns were found.
[Indirect Prompt Injection] (LOW): The skill contains templates that ingest untrusted user data into AI prompts, creating a vulnerability surface. Specifically, the prompt enhancement example in templates/ai-image-generation.ts lacks input sanitization.
Ingestion points: templates/ai-image-generation.ts (via userPrompt in the /generate/enhanced endpoint) and templates/ai-gateway-integration.ts (via prompt).
Boundary markers: Absent in the prompt templates.
Capability inventory: AI.run (inference) and R2Bucket.put (file storage).
Sanitization: The templates do not demonstrate input validation or escaping for the ingested prompts.

cloudflare-workers-ai