Vector has added AI agents and Model Context Protocol (MCP) capabilities to its CANoe development and testing environment, letting engineers automate development, analysis, and validation workflows with natural-language prompts. The functionality, available through CANoe 20 SP2 and the CANoe AI Package, turns a written requirement into a CAPL test that the agent executes, debugs, and reruns automatically.
Vector said workflows that previously took hours or days can now be completed within minutes. Engineers keep control over how much autonomy the agents have, and every operation stays visible inside the development environment.
How the AI agent workflow works
Users provide a requirement as a prompt, and the AI agent generates the matching CAPL test, executes it in CANoe, analyzes failures, modifies the code, and reruns the test. The engineer reviews the resulting scenario through synchronized CANoe windows before approving the results.
The system runs on an open AI architecture made up of agents, skills, and MCP tools. Users can bring their own large language model, including models behind services such as GitHub Copilot or Claude. Vector provides the AI integration layer that connects those models to CANoe's functionality.
The MCP-based approach also lets engineers read and modify configurations, control simulations, create tests, analyze communication flows, and generate or optimize CAPL, C# and Python code through prompts. Vector's documentation is available to the agents through Vector-RAG, a retrieval-augmented generation system that grounds responses in verified technical information rather than relying on language-model assumptions.
Vector said: "From a single prompt, an AI agent creates the appropriate CAPL test from a requirement, executes the test in CANoe, analyzes errors, corrects the CAPL code, and reruns the test."
Architecture and availability
The design supports both new and experienced CANoe users, from straightforward queries to fully automated workflows that combine multiple development and testing activities. The MCP Server integrated into CANoe is included in Version 20 SP2.
The CANoe AI Package is available as a free download and works with CANoe 20 SP2 and later versions. Users must supply their own language model. For developers working with AI automation in test environments, the AI for IT & Development resources cover similar ground, and the MCP (Model Context Protocol) tag tracks developments in this protocol.
Why this matters for IT and development teams
For engineers who write and maintain test scripts, this shifts the work from writing code line by line to reviewing code an agent produced. The practical gain is speed: a test that took a day to build and debug can be generated and validated in minutes. The control mechanism matters too - the agent's work is reviewable before approval, so teams can adopt the workflow without handing over decision-making.
Your membership also unlocks:
