MalveonChecks whether your AI coding agent really built what your plan and your prompts asked for, by reading the code, never by asking the agent. What you see
Getting started
New run (the + button, also on the results page) is for starting a new task: one click starts a fresh session, so earlier prompts and changes stop counting, then asks which plan files this run is about and checks. Use it when your agent starts on something new. Run with build and tests also runs your project's own build, lint, type check and tests. It's a separate button because it runs real commands and can take a few minutes. Your agent can use it tooThe extension connects malveon to your editor's own agent automatically (tested in VS Code; Cursor has its own connection, not yet tested there), so the agent can run the check itself and read the same result you see. For an agent in a terminal or another editor, use Copy setup for another agent in the sidebar menu. You stay in chargeAnything that changes a setting or stops a prompt from being checked asks you first, in a dialog. The agent can't answer those for you. The same goes for running a project's own build and test commands: the first time, malveon shows you exactly which commands would run and asks, once per project. A repo you just cloned can't make malveon run its scripts on its own. PrivacyMalveon runs on your machine and reads your project there. Your code, file names, plans and prompts never leave it; your prompts are saved only in your project, in Two things do go over the network:
LicenseFree to use during the beta, for personal and commercial work. See the license included with the extension. |