etherscan-mcp-skill
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were identified in the skill files.
- [EXTERNAL_DOWNLOADS]: The skill communicates with the official Etherscan MCP endpoint at
https://mcp.etherscan.io/mcp. This is a recognized service for on-chain data retrieval. - [CREDENTIALS_SAFE]: The skill uses
uxc authto manage theETHERSCAN_API_KEY. It explicitly instructions the user to use environment variables or a 1Password secret manager (op://...), adhering to secret management best practices. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from the Ethereum blockchain, including contract source code and transaction details. This constitutes an indirect prompt injection surface. The risk is minimized because the skill is intended for investigative analysis, encourages parsing the JSON envelope first, and lacks automated execution logic that would act on the retrieved data without user oversight. 1. Ingestion points: blockchain transaction data and contract source code via the Etherscan MCP endpoint. 2. Boundary markers: the instructions specify parsing the JSON envelope (
ok,kind,protocol,data). 3. Capability inventory: theuxcCLI is used to execute defined tools; no arbitrary command execution or file writes on retrieved data are documented. 4. Sanitization: reliance on structured JSON output for automation and manual review of text content.
Audit Metadata