web3-testing
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides templates that ingest smart contract source code and data from external blockchain RPC providers, creating a potential surface for indirect instructions.
- Ingestion points: Reads local smart contract source files and processes data returned from Ethereum RPC providers during mainnet forking.
- Boundary markers: The templates do not define explicit delimiters to separate untrusted contract data from the agent's instructional context.
- Capability inventory: The skill uses Hardhat and Foundry CLI tools for testing and interacts with the Etherscan API for contract verification.
- Sanitization: Standard development libraries are used, but no specific sanitization logic is provided to handle potentially malicious instructions embedded in external contract code or RPC responses.
- [SAFE]: The skill demonstrates safe credential management by utilizing environment variables (
process.env) and explicitly instructing users to keep RPC credentials in a secret store. It also includes warnings against executing unverified scripts and deploying to funded networks during testing.
Audit Metadata