web3-identity-auth

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through ENS text records. The ENSResolverTool in scripts/ens_resolver.py fetches user-controlled data such as descriptions and URLs from the Ethereum Name Service. Because the agent context defined in SKILL.md does not utilize boundary markers to separate this untrusted data from instructions, a malicious ENS record could be used to influence the agent's behavior.
  • Ingestion points: Text records retrieved in scripts/ens_resolver.py and interpolated into the AuthenticatedAgent system prompt in SKILL.md.
  • Boundary markers: Absent from the system prompt template.
  • Capability inventory: The agent has access to ENSResolverTool and potentially sensitive WalletTool functionality.
  • Sanitization: No sanitization or escaping is applied to external data retrieved from ENS resolvers before it is provided to the language model.
  • [EXTERNAL_DOWNLOADS]: The skill performs network requests to api.thegraph.com to facilitate batch resolution of ENS names. This is an expected operation using a well-known service in the blockchain ecosystem and is consistent with the skill's primary purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 07:33 PM