Skip to content

Contributing

Review this guide to understand the core design philosophy, alignment goals, and contribution guidelines before initiating work on the codebase.

Project Scope

Rivet is a hard fork of Task. Rivet diverges from the upstream project to serve a specific purpose: to act as a minimal, lightweight task runner engine built strictly for speed and simplicity.

To maintain this focus, the project adheres to the following principles:

  • Hyper-Minimalism: Feature bloat is avoided. Code that unnecessarily expands the core engine footprint will not be merged.
  • Strict Alignment: Only contributions explicitly aligned with the core goals of Rivet are accepted.

Rivet vs. Upstream

Info: Rivet is a hard fork of the upstream project, the codebase has been significantly altered.

Because Rivet has a highly restricted, niche scope, standard or broad feature requests are generally not a fit for this repository.

  • General-purpose utility or broad features: Contribution to the upstream Task project is advised, given its larger, more established community.
  • Optimization of the minimal footprint: Proposals that directly enhance the speed or simplicity of the core engine belong in this repository.

Introducing New Concepts

Creative extensions are welcome, but the core runner engine must remain isolated.

  • Separate Ecosystem Repositories: Major new concepts—such as a Terminal User Interface (TUI)—must be developed as standalone projects in a dedicated repository rather than integrated into the main engine codebase.
  • Awareness of Effort: Building a new sub-project requires a long-term commitment to maintenance, issue triage, and updates. Proponents of new ecosystem tools must be prepared to sustain that effort.

Technical Expectations & AI Policy

Rivet is engineered in Go with a focus on high performance and structural integrity. Because the engine requires precise logic, strict standards apply to code contributors:

  • Sufficient Experience & Skill: Technical proficiency in Go, concurrent programming, and CLI architecture is required. Code must be clean, idiomatic, and fully tested. E2E tests are preferred over Go unit tests.
  • Core Codebase AI Usage: The core codebase leverages AI-assisted development tools to accelerate engineering. However, unverified AI output or entirely automated code generation is not accepted. E2E tests are preferred as the validation method.
  • No AI-Authored Pull Requests: Do not bother opening pull requests with AI-generated code. Instead, simply open an issue, preferably with an E2E test that can become part of the regression test suite.
  • No AI-Authored Unit Tests: Pull requests containing AI-generated test cases will be closed.