cloudflare
Warn
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The documentation for the Sandbox and Containers modules provides patterns for executing arbitrary shell commands using
sandbox.exec()and defining container entrypoints. This allows for arbitrary code execution within isolated environments. - [DATA_EXFILTRATION]: The Tail Workers module includes instructions on how to use the
getUnredacted()method to bypass default security filters. This enables the extraction of sensitive request data, including authentication headers, tokens, and cookies, which are normally hidden for security. - [DATA_EXFILTRATION]: The skill documents multiple pathways for data extraction, including the Pipelines and Queues APIs, which allow sending data to external sinks or storage buckets.
- [CREDENTIALS_UNSAFE]: Many implementation patterns involve the management of powerful account-level API tokens, service keys, and database credentials. While the skill recommends secret management tools, the patterns often involve handling these sensitive values in environment variables.
- [PROMPT_INJECTION]: The Agents SDK and Email Workers modules process untrusted data from external sources such as emails and WebSocket messages. The provided instructions lack clear boundaries or sanitization patterns to prevent this data from influencing AI model behavior (Indirect Prompt Injection).
Audit Metadata