OpenZeppelin MCP Agent
An AI agent for smart contract security auditing launched by OpenZeppelin, built on the MCP protocol. It exposes contract security auditing capabilities as a standardized MCP tool server, enabling AI models to invoke contract security checks via typed JSON Schema. The agent incorporates a two-phase verification mechanism.
Timeline (last 90 days)
OpenZeppelin MCP Agent combines security auditing capabilities with the Model Context Protocol, enabling AI models to invoke contract security checking tools in a standardized manner.
The two-phase dry run mechanism allows the model to first perform a dry run to verify logic and parameter correctness before entering the actual execution phase.
The method for evaluating whether a model calls a tool is to run a fixed prompt panel, record whether the model mentions or invokes a specific tool, and repeat the test after each model version update.
All Facts (3)
OpenZeppelin MCP Agent combines security auditing capabilities with the Model Context Protocol, enabling AI models to invoke contract security checking tools in a standardized manner.
50%UnverifiedThe two-phase dry run mechanism allows the model to first perform a dry run to verify logic and parameter correctness before entering the actual execution phase.
50%UnverifiedThe method for evaluating whether a model calls a tool is to run a fixed prompt panel, record whether the model mentions or invokes a specific tool, and repeat the test after each model version update.
50%