erc-8004
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill fetches data from external IPFS and HTTPS sources.
scripts/get-agent.shretrieves agent profile JSON from IPFS gateways (gateway.pinata.cloud, ipfs.io) or user-defined HTTPS URLs found on-chain.- COMMAND_EXECUTION (LOW): The skill executes blockchain transactions and data encoding scripts.
- Multiple scripts use the
bankrCLI to prompt the user for transaction signing and submission. node -eis used throughout the scripts to perform hex encoding and decoding for Ethereum ABI parameters.- PROMPT_INJECTION (LOW): Surface for indirect prompt injection exists when fetching untrusted agent profiles.
- Ingestion points:
scripts/get-agent.sh(fetches remote JSON profiles). - Boundary markers: Absent; the raw or jq-parsed JSON is output to the agent context.
- Capability inventory: Significant; the agent can trigger ETH bridging and on-chain registration transactions.
- Sanitization: Partial; the script validates the URI starts with
ipfs://orhttps://before fetching, but doesn't sanitize the returned JSON content.
Audit Metadata