frontend-apps
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides patterns for fetching and displaying untrusted content from the Sui blockchain (such as NFT metadata and object fields), which creates a surface for indirect prompt injection. Malicious on-chain data could contain instructions designed to influence an agent's behavior during data processing.
- Ingestion points: Data enters the agent's context through Core API methods like
client.core.listOwnedObjects,client.core.listBalances, andgetObject(referenced inqueries.mdandtransactions.md). - Boundary markers: The provided code snippets for rendering data do not demonstrate the use of delimiters or specific instructions to disregard embedded commands in the processed data.
- Capability inventory: The skill grants the ability to perform high-stakes operations including
signAndExecuteTransaction(which can move assets) andsignPersonalMessage(used for authentication). - Sanitization: No explicit sanitization or filtering of external blockchain content is shown in the provided UI examples.
Audit Metadata