Per-repo documentation — each repo's docs/ and README, ingested and associated with the repo. Rendered through the <<<RepoDocs>>> tag.
← source docs · codebean_mcp/README.md
An MCP that lets an LLM author code as an in-memory model — packages, classes, functions — without touching files or git. The model is the "bean"; save() materializes + commits behind the scenes. The point is the focused sub-mind: it works on code structure and reads it via render(), never distracted by disk, paths, or branches.
- new_module(name) - add_class(module, name, bases, doc) - add_function(target, name, params, body, doc, decorators) — target = module or module.Class (self inserted for methods) - set_body(target, body) — module.func or module.Class.method - render(module?) — read the generated source - tree() — structure (modules → classes/methods/functions) - save(message) — persist (materialize + git commit; git is concealed — you just describe the change)
Config: CODEBEAN_ROOT (hidden workspace, default ~/.codebean/work), CODEBEAN_PROJECT. Generated source is valid Python (verified by compile() in tests).