skill-security-framing

To prevent prompt injection when analyzing external content, you must wrap untrusted data in a secure framing context before processing. This isolates external material, ensuring safe analysis and preventing unintended execution of malicious instructions.

Security framing for untrusted content is a workflow that wraps external data in an isolated context to prevent data leakage. It enforces URL validation and content wrapping before analysis, ensuring external material cannot trigger unintended execution.

To secure webhook payloads, apply a security framing workflow that validates the source and wraps the payload in an untrusted context. This prevents data leakage and prompt injection before you process the API responses.

Yes, this security framing approach works for fetched documents and API responses by applying stepwise rules. It enforces URL validation and wraps all external material in an untrusted context to ensure sanitized outputs across workflows.

Without a security framing context, analyzing external content risks prompt injection, data leakage, and unintended execution. Bypassing URL validation and content wrapping exposes your workflow to malicious instructions embedded in untrusted material.