erc8004-on-abstract

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill defines processes for discovering agents and reading reputation, which involves ingesting natural language data from external JSON files (agentURI and feedbackURI).
  • Ingestion points: External metadata files hosted on HTTPS or IPFS, as described in references/agent-uri-schema.md, contain fields like name and description which are processed by the agent.
  • Boundary markers: The provided schemas and documentation do not implement delimiters or instructions to ignore potential commands embedded in the fetched metadata.
  • Capability inventory: The skill includes blockchain interaction capabilities (contract reads/writes) via the viem library; it does not include arbitrary shell command execution, file system writes, or general-purpose network exfiltration tools.
  • Sanitization: No sanitization or validation logic is provided for the content of natural language fields retrieved from the registries.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 09:00 AM