# reaktiv > Signals for Python: reactive, declarative state management with automatic dependency tracking. Signals hold state, computed values derive from them lazily, effects run side effects, and resources load async data from reactive parameters. Read the agent guide first: it states the mental model and the rules for writing reaktiv code. The installed package contains the same guide, matching its version, at `reaktiv/AGENTS.md`. ## Guide - [Agent guide](https://raw.githubusercontent.com/buiapp/reaktiv/main/src/reaktiv/AGENTS.md): mental model, which primitive to use, rules, patterns and pitfalls - [Core concepts](https://reaktiv.bui.app/docs/core-concepts/): signals, computed values, effects and dependency tracking - [Reactive models](https://reaktiv.bui.app/docs/api/reactive-model/): grouping state, derived values, effects and resources in a class - [Resource guide](https://reaktiv.bui.app/docs/resource-guide/): async loading, status, cancellation ## Optional - [Advanced features](https://reaktiv.bui.app/docs/advanced-features/): custom equality, batching, threads and event loops - [subtle](https://reaktiv.bui.app/docs/api/subtle/): Watcher and primitives for framework authors - [Changelog](https://raw.githubusercontent.com/buiapp/reaktiv/main/CHANGELOG.md)