skills/meshjs/skills/mesh-wallet/Gen Agent Trust Hub

mesh-wallet

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external Cardano transaction data and authentication messages, which presents a surface for indirect prompt injection.\n
  • Ingestion points: Methods such as signTx, signData, and submitTx described in WALLET.md and SKILL.md are designed to receive external inputs like transaction hex strings and data messages.\n
  • Boundary markers: The skill does not provide specific instructions or delimiters to isolate these external inputs from the agent's instruction context.\n
  • Capability inventory: The skill leverages the MeshJS SDK to perform network queries (e.g., fetching UTxOs and balances) and network writes (e.g., submitting signed transactions).\n
  • Sanitization: No explicit validation or sanitization logic for transaction hexes or message strings is included in the provided implementation patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 03:09 PM
Security Audit — agent-trust-hub — mesh-wallet