building-ai-agent-on-cloudflare
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: Vulnerability surface for indirect prompt injection via untrusted data ingestion.
- Ingestion points: Processes user input in onMessage handlers, external tool responses in handleChatWithTools, and retrieved document context in RAGAgent.
- Boundary markers: The prompt templates lack explicit delimiters or instructions to ignore embedded commands in the processed data.
- Capability inventory: Includes executable SQL queries via this.sql, network requests using fetch, and automated task scheduling via this.schedule.
- Sanitization: SQL queries are correctly parameterized using tagged templates, but no sanitization is applied to data interpolated into AI prompts.
- [EXTERNAL_DOWNLOADS]: Fetches starter templates and reference implementations from official Cloudflare resources.
- Evidence: Utilizes the npm create cloudflare@latest command to download the cloudflare/agents-starter template.
- Evidence: References example implementations within the github.com/cloudflare/agents repository.
Audit Metadata