cloudflare
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external Cloudflare API responses, such as log entries and build status messages, while maintaining capabilities for sensitive operations. * Ingestion points: Tool outputs from execute and search containing external logs, DNS records, and build metadata as described in SKILL.md and references/workflows.md. * Boundary markers: No specific instructions are provided for the agent to use delimiters or markers to isolate external content from instructions. * Capability inventory: The execute tool can perform sensitive actions such as modifying DNS records, deploying Worker scripts, and deleting R2 storage buckets. * Sanitization: The skill includes extensive sanitization logic in references/safety-and-permissions.md that requires redacting secrets and obtaining explicit user consent before write operations.
- [DYNAMIC_EXECUTION]: The skill requires the agent to generate and execute small JavaScript code snippets to interact with the Cloudflare API. * Evidence: SKILL.md specifies that the execute tool is used to call Cloudflare APIs with minimal JavaScript produced by the agent.
Audit Metadata