iblai-vibe-agent-create
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The client-side helper in assets/admin-client.ts.j2 retrieves a platform authentication token (dm_token) from localStorage. This token is subsequently forwarded in the Authorization header for all administrative API calls. Storing sensitive authentication tokens in localStorage is a common but risky practice, as the data is accessible to any script within the same origin, making it vulnerable to extraction if the application suffers from Cross-Site Scripting (XSS).
- [INDIRECT_PROMPT_INJECTION]: The setup interface in assets/setup-screen.tsx.j2 allows administrators to provide an agent name and purpose which is directly mapped to the system_prompt field in the platform API request in assets/admin-agents-route.ts.j2.
- Ingestion points: assets/setup-screen.tsx.j2 (inputs for newName and newPurpose).
- Boundary markers: None. User-supplied strings are interpolated directly into the JSON request body.
- Capability inventory: The skill facilitates the creation of new AI agents with administrative privileges on the ibl.ai platform.
- Sanitization: No validation or prompt-specific sanitization is applied to the input purpose string before it is deployed as the agent's system prompt.
Audit Metadata