single-sign-in
Pass
Audited by Gen Agent Trust Hub on Nov 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references the
@icp-sdk/authpackage and@dfinity/static-site. These are official resources from the vendor and are necessary for the described functionality. - [INDIRECT_PROMPT_INJECTION]: The skill implements a redirection mechanism that relies on external input via URL parameters.
- Ingestion points: The code extracts a
nextparameter fromlocation.searchwithin the/reauthroute logic inSKILL.md. - Boundary markers: Absent. The implementation does not include logic to validate that the redirect target is a relative path or belongs to an authorized domain.
- Capability inventory: The skill uses
location.replace()to navigate the user to the provided URL and passes the value to theauthClient.signInmethod. - Sanitization: No validation or sanitization is applied to the
nextparameter before it is used for client-side redirection. While common in authentication flows, this creates a potential open-redirect surface if the parameter contains an external malicious URL.
Audit Metadata