basescope is a read-only security layer for on-chain AI agents, providing AI assistants (such as Claude, Cursor, and any MCP client) with secure, read-only on-chain capabilities on Base and other EVM chains in the form of a Model Context Protocol (MCP) server. Its core task is to answer one question before any wallet signs: "Is this token / contract / approval actually safe?"
Purpose basescope is used for on-chain security checks before a transaction or approval occurs. It checks for honeypots, rug-pull permissions, hidden taxes, malicious addresses, and risky approvals, and in the process queries balances, ENS names, Basenames (*.base.eth), verified source code, Gas, and prices. It is positioned as a "checker, not a wallet," a safety companion for agents with transaction capabilities.
Main capabilities
- Read-only design: no code path can sign or send transactions, and no private key can be configured, so funds cannot be moved.
- No API key required: works out of the box, based on free public data sources; optional keys are only for future rate-limit increases and are not required.
- Base-first, multi-chain ready: tuned for Base, while also supporting Ethereum, Optimism, Arbitrum, and Polygon.
- Cross-validation: token safety results cross-reference two independent sources (GoPlus + honeypot.is) and label their provenance, making it easy for agents to spot inconsistencies between the two.
- Address resolution: address fields accept 0x addresses, ENS names, or Basenames (*.base.eth); Basenames are automatically resolved via the Base L2 Resolver.
- Chain parameter: chain accepts base (default), ethereum, optimism, arbitrum, or polygon.
How to use Add basescope to any MCP client, such as Claude Desktop (claude_desktop_config.json), Cursor (.cursor/mcp.json), or another MCP client. No installation step is required; npx fetches it on first run. After restarting the client, you can ask questions like "Use basescope to check if token 0x… on Base is safe.". You can also set the server command to node with the argument ["/absolute/path/to/basescope/dist/index.js"].
Configuration (all optional) Everything works without configuration. For more frequent use (public RPCs have rate limits), you can set your own RPC endpoint.
Open-source/free services it depends on On-chain reads use viem + public RPC; token and address security use GoPlus Security and honeypot.is; verified source code uses Sourcify; prices use DefiLlama.
Use cases Suitable for AI agents performing security checks before executing transactions or approvals on Base and EVM chains, such as determining whether a token is a honeypot, whether a contract has rug-pull permissions, whether an approval is risky, and querying balances, ENS/Basenames, verified source code, Gas, and prices.
Limitations basescope reports heuristic signals, not guarantees, and does not constitute financial advice. Security APIs can be wrong or outdated, and a "low risk" result does not promise safety. Do your own research before trading. On Base (OP-Stack L2), get_gas only reports L2 fees; the total cost also includes L1 data fees. Features mentioned in the materials but not expanded on include: transaction/calldata simulation and decoding ("what would this actually do?"), and an optional Etherscan V2 fallback for source code and ABI/proxy resolution.
