tuicr

Product Introduction

1. Definition: tuicr is a terminal user interface (TUI) application for conducting code reviews. It is a single static binary built in Rust that renders GitHub-style diffs and enables a full PR review workflow directly within the terminal environment.
2. Core Value Proposition: tuicr exists to eliminate the disruptive context-switching between a developer's terminal/editor and a web browser during the code review process. It provides a faster, keyboard-driven workflow for engineers who primarily operate in the terminal, integrating review capabilities directly into their existing development flow.

Main Features

1. GitHub-Style Diff Scrolling: tuicr presents all changed files in a repository as one continuous, scrollable view, mimicking the familiar GitHub pull request diff interface. This is powered by its internal diff parsing engine, which works across Git, jj, and Mercurial (hg) version control systems, auto-detecting the VCS in use.
2. Vim-Keybinding Navigation & Modal Editing: The TUI implements a comprehensive vim keybinding scheme, not just basic navigation. This includes modal interaction, visual line/block selection (with v/V), search (/), jumps to specific lines ({N}G), and viewport scrolling (Ctrl-d/Ctrl-u). It provides a native feel for vim and neovim users.
3. Structured, Multi-Level Commenting: Users can create PR-style comments at four distinct levels: on a single line, on a selected range of lines, on an entire file, or on the overall review. Each comment can be classified by type (e.g., Issue, Suggestion, Note, Praise). The system uses visual mode selection to define comment ranges precisely.
4. Multi-Target Review Export: Completed reviews can be exported through three distinct pathways. The :submit command pushes inline comments and a review verdict (Approve, Comment, Request Changes) directly to the live GitHub Pull Request via the authenticated GitHub CLI (gh). The y or :clip command copies a structured markdown summary to the clipboard for LLM/agent use, and the --stdout flag pipes the same markdown to standard output for scripting.
5. Agent & Automation Integration: tuicr is designed for integration with AI coding assistants. It ships with a dedicated "skill" for Claude Code/Codex that allows an agent to open tuicr in a tmux split. The structured markdown output is specifically formatted to be parsed by LLMs, creating a closed-loop feedback system between developer and agent.

Problems Solved

1. Pain Point: The inefficient workflow of constantly toggling between a local terminal (for checking out code, running tests) and a web browser (for viewing diffs and typing comments) during code review, which breaks focus and slows down the review cycle.
2. Target Audience: The primary user personas are backend/system developers, DevOps engineers, and open-source maintainers who are terminal-centric, proficient with vim keybindings, and work extensively with Git, jj, or Mercurial repositories. A secondary persona includes developers using AI coding agents who need a structured way to provide batch feedback.
3. Use Cases: Essential for reviewing complex PRs locally before pushing feedback; for conducting deep, focused code analysis without browser distractions; for reviewing patches in jj or Mercurial workflows which have less web tooling; and for generating actionable, line-specific feedback blocks to provide to an AI programming assistant.

Unique Advantages

1. Differentiation: Unlike web-based tools (GitHub/GitLab UI) or other TUI diff viewers (e.g., lumen), tuicr uniquely combines full vim emulation, inline comment creation, and the ability to push those inline comments back to GitHub as a native review. It also supports jj and Mercurial, which most cloud-centric tools ignore. Compared to gh pr review, it supports true line-level commenting.
2. Key Innovation: Its three-pronged export mechanism is a key innovation. It bridges the gap between local review tools and platform APIs (:submit), human-AI collaboration (y/:clip), and pure automation (--stdout). This makes it both a daily driver for developers and a composable component in modern, toolchain-driven development workflows.

Frequently Asked Questions (FAQ)

1. How does tuicr push comments to GitHub? tuicr uses the officially authenticated GitHub CLI (gh) to submit reviews. The :submit command opens a picker for the review verdict, then formats and posts all inline comments to the correct lines on the live pull request, creating a native GitHub review.
2. Can I use tuicr with AI coding agents like Claude Code? Yes, tuicr is explicitly designed for this. It offers a /tuicr skill bundle for Claude. The agent can launch tuicr, and the developer's comments are output as structured, numbered markdown that the agent can easily parse and act upon, creating an integrated review loop.
3. What version control systems does tuicr support? tuicr auto-detects and supports Git, jj (which is Git-backed), and Mercurial (hg). It is a single static binary with no runtime dependencies, making it equally portable across all three VCS environments.
4. Is tuicr suitable for non-vim users? While its power user features are heavily optimized for vim keybindings, basic navigation (arrow keys, simple search) is possible. However, its core value proposition is strongest for developers already proficient in vim's modal editing paradigm.
5. How do I install tuicr on my system? tuicr can be installed via a one-line shell script (curl -fsSL tuicr.dev/install.sh | sh), through package managers like Homebrew (brew install agavra/tap/tuicr) and mise, via Cargo (cargo install tuicr), or run directly with Nix (nix run github:agavra/tuicr).