skills/ethpandaops/ai-cookbook/eip/Gen Agent Trust Hub

eip

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to perform git clone operations on Ethereum specification and client repositories. It also utilizes grep and git log to search for EIP references within the downloaded codebases and commit history.
  • [EXTERNAL_DOWNLOADS]: Downloads source code from multiple public GitHub repositories, including official Ethereum specifications and major consensus and execution layer client implementations such as Geth, Prysm, and Lighthouse. These references target well-known open-source projects within the Ethereum ecosystem.
  • [DATA_EXFILTRATION]: Performs network requests using WebFetch and WebSearch to retrieve information from eips.ethereum.org, forkcast.org, and ethresear.ch. These operations are consistent with the skill's primary purpose of aggregating Ethereum-related technical data.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it ingests untrusted data from external web sources (EIP content and forum posts) and processes it using the agent's reasoning capabilities.
  • Ingestion points: Data retrieved via WebFetch from external websites and repositories.
  • Boundary markers: None identified; the instructions do not include specific delimiters or warnings to ignore instructions embedded in the external content.
  • Capability inventory: The skill has access to powerful tools including Bash, Task, and file system utilities.
  • Sanitization: No explicit sanitization or filtering of external content is performed before interpolation into the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 08:56 PM
Security Audit — agent-trust-hub — eip