port
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted web codebase files including manifests, routing configurations, and source code to generate detailed porting blueprints. These blueprints are subsequently passed to downstream implementation agents such as 'Native' or 'Scaffold'. There is a possibility that malicious instructions embedded within the analyzed web codebase (for example, in code comments, metadata, or documentation) could be included in the generated specifications, potentially influencing the behavior of downstream agents.
- Ingestion points: The skill reads web application manifests, routing configs, state stores, API clients, storage usage, and auth flows from the provided codebase.
- Boundary markers: The skill does not explicitly define delimiters for surveyed content in its output to downstream agents.
- Capability inventory: The skill generates detailed implementation specifications, database schemas, and API contracts for other agents to execute.
- Sanitization: No explicit sanitization or escaping of extracted web content is defined before it is placed into the output blueprints.
Audit Metadata