# valgebra-planning > Plans of tool calls for agents, proven well-typed for every output the tool contracts admit. Plans of tool calls for agents, proven well-typed for every output the tool contracts admit. A tool's MCP inputSchema and outputSchema are read by valgebra-jsonschema into valgebra's sets; a program of calls and type tests is accepted only where every value it passes on is proven inside the argument it fills. A program written elsewhere, by hand or by a language model, is verified the same way, link by link. ## Getting started - [Home](): What valgebra-planning is, how it rests on valgebra and valgebra-jsonschema, and where to start. - [The command line](): Plan a library's tasks, critique a program, run one against a live server, or serve the proxy, from the command line. - [Code mode](): A model's program proven before it runs and run so the proof holds, and the two kinds of check the proxy holds. - [Recipes](): Tasks done end to end over tool listings in hand -- a model's program checked before it runs, a server answering outside its contract, an approval a policy requires, a notice held to the address on file, a privilege update, and a policy that cannot fire. ## Release notes - [Release notes](): Every released version, what it changed, and what counts as a break. ## Explanation - [How it works](): What a program of tool calls is, what proving it means, and how valgebra-planning, valgebra-jsonschema and valgebra share the work. - [The method](): Contracts, the cells, the planner, the verifier, the rules a program is held to, the critic, and the limits. - [Results](): The studies, each count with the command that checks it. ## Reference - [Python API](): The public modules, the names each promises, the two needing the mcp extra, and their reference read from the docstrings. - [Works cited](): Every work the method and the modules cite, under its key.