cloudflare
Pass
Audited by Gen Agent Trust Hub on Apr 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Uses
curlandjqto interact with the Cloudflare API and process responses. - [EXTERNAL_DOWNLOADS]: Connects to
api.cloudflare.comto fetch zone and account data. - [DATA_EXFILTRATION]: Transmits user-provided API tokens to Cloudflare's API endpoints for authentication and resource management.
- [PROMPT_INJECTION]: Ingests and processes untrusted data from external API responses, which represents an indirect prompt injection surface.
- Ingestion points: DNS record contents, zone metadata, and Worker script listings retrieved from Cloudflare.
- Boundary markers: Not used; API responses are incorporated directly into the agent's context.
- Capability inventory: Includes network access via
curland temporary file writing to/tmp/cloudflare_request.json. - Sanitization: No evidence of validation or sanitization of external data before it is processed.
Audit Metadata