rigup.nix

rigup is a comprehensive system designed for creating self-contained and modular configurations for AI agents. By utilizing the Nix package manager, it allows developers to package knowledge, instructions, and technical skills as 'riglets'—executable units of logic and context. This platform addresses the challenge of managing complex agent environments by treating infrastructure and operational knowledge as code, ensuring that your agent is always equipped with the specific tools and documentation required for the task at hand. It is ideal for engineering teams and power users who need reproducible, version-controlled agent setups that can be shared across projects or integrated with existing Claude-compatible skills.

• Enables the creation of modular 'riglets' that bundle metadata, process instructions, and required nix packages into a single unit.

• Provides a 'knowledge-first' architecture, where documentation is treated as a primary payload that can be lazily or eagerly disclosed to the agent during runtime.

• Features a powerful CLI tool (rigup) for inspecting, building, and running agent environments, including subshell support and automated manifest generation (RIG.md).

• Offers fine-grained control over module evaluation, including advanced handling of lib.evalModules, priority management (mkDefault/mkForce), and complex dependency ordering.

• Supports high-level interoperability with external ecosystems like the Claude Skills marketplace, facilitating the reuse and adaptation of third-party agent capabilities.

• Use the CLI to scaffold new projects via templates, ensuring that environment setup and dependency management are handled correctly from day one.

• Leverage the module system to define 'entrypoints' that wrap specific agent harnesses, allowing for seamless transition between development, shell-based usage, and production-like execution.

• Consult the embedded technical documentation for navigating the 'dark corners' of the Nix module system, such as handling infinite recursion, managing module deduplication via keys, and utilizing freeformTypes for flexible configuration.

• The system works by consuming nix flakes as inputs; ensure that your environment is configured for Nix with flakes enabled to benefit from automated binary caching via providers like garnix.io.

• Be aware that the system relies on deep familiarity with Nix expression language; while it provides high power and flexibility, it is best suited for users already comfortable with functional infrastructure-as-code paradigms.