sui-jsonrpc-skill
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill targets the official Sui mainnet fullnode (
https://fullnode.mainnet.sui.io), which is a well-known and trusted blockchain service. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external RPC endpoints, which is a standard surface for indirect prompt injection. However, it includes guardrails to focus on structured JSON fields and read-only operations.
- Ingestion points: JSON-RPC responses from Sui fullnode endpoints (SKILL.md, references/usage-patterns.md).
- Boundary markers: Instructions advise parsing specific stable fields (
ok,kind,protocol,data,error) from the JSON output (SKILL.md). - Capability inventory: The skill uses
uxcto perform network requests and manage local subscriptions (SKILL.md). - Sanitization: The skill recommends keeping automation on the JSON output envelope rather than raw text (SKILL.md).
Audit Metadata