authentication-authorization
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's metadata contains an installation command that downloads the instruction file from a GitHub repository (splits-network/splits). This is a standard mechanism for agent skill distribution.\n- [INDIRECT_PROMPT_INJECTION]: The skill defines patterns for handling authentication identifiers and search filters, which represent an ingestion surface for potentially untrusted data.\n
- Ingestion points: clerkUserId from headers and filters in repository queries.\n
- Boundary markers: Emphasizes identity resolution via resolveAccessContext and warns against trusting client-controlled headers.\n
- Capability inventory: Performs database operations using the Supabase client.\n
- Sanitization: Relies on third-party libraries (Clerk, Supabase) for JWT verification and query parameterization.\n- [SAFE]: No malicious patterns such as credential exfiltration, prompt injection, or obfuscation were detected. The skill provides legitimate architectural guidance for secure application development.
Audit Metadata