cloudflare
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Documents the use of official Cloudflare SDKs and well-known development libraries such as
@cloudflare/wrangler,@cloudflare/puppeteer,zod,drizzle-orm, andpgthroughout its configuration and API guides. - [COMMAND_EXECUTION]: Provides numerous examples of shell commands for resource management via
wranglerandcurl. It also describes thesandbox.execmethod for running code within isolated containers, which is an intended feature of the Cloudflare Sandbox SDK. - [INDIRECT_PROMPT_INJECTION]: As a documentation repository, the skill represents a large volume of data that instructs the AI agent on how to interact with services.
- Ingestion points: Extensive markdown files in the
references/directory describing Cloudflare products. - Boundary markers: Not present; the documentation uses standard text formatting without specific delimiters for LLM instructions.
- Capability inventory: documents capabilities for shell execution, container management, and network API calls.
- Sanitization: Not present; however, the content is instructional and intended for the agent to follow in its primary role as a Cloudflare platform guide.
Audit Metadata