Hunk
Review-first terminal diff viewer for inspecting agent-authored changesets with multi-file navigation, inline AI notes, watch mode, and Git/Jujutsu/Sapling support.
Do not bounce yet
Read the fit check, compare one alternative, then decide whether the vendor page is still your best next click.
Quick Verdict
Fast fit check before you leave the page
Make the fit call first. Vendor pages are good at selling, but they rarely tell you where the product is a bad match.
- Developers reviewing large local diffs from Claude Code, Codex, OpenCode, or similar agents
- Terminal-heavy engineers who want a better alternative to raw git diff for AI-generated changes
- Teams building review gates around vibe-coded pull requests
- Hunk does not replace a coding agent or hosted pull-request reviewer; it is the local diff-review layer in the workflow.
- Teams still need discipline around what the agent annotates and what humans verify, because inline AI notes can be wrong just like chat output.
- The strongest fit is terminal-heavy review; developers who prefer browser or IDE review surfaces may not adopt it as their default.
Compare Next
Take one more internal step before the vendor pitch
This is where visitors usually jump out too early. Read one deeper take or open one alternative so the next click is informed instead of impulsive.
Alternative profile
CodeRabbit
AI code review platform for pull requests, IDEs, and the CLI, built to catch bugs and keep fast-moving agentic teams from merging sloppy changes.
Alternative profile
Open Code Review
Open-source AI code review CLI that reads diffs, explores repo context, and produces line-level review comments for agent-heavy pull requests.
Alternative profile
Every Code
Community Codex CLI fork that adds Auto Drive orchestration, background Auto Review, browser/CDP integration, theming, and multi-agent commands.
Hunk is worth listing because agentic coding has made local diff review harder, not easier. Coding agents can produce large, fast-moving changesets, and a raw git diff in a pager is a weak review surface for that workload. Hunk turns those changes into a navigable terminal review stream with inline notes, watch mode, and a live-session workflow that agents can interact with.
Hunk is a terminal diff viewer built for the review bottleneck created by coding agents. Instead of dumping a raw git diff into a pager, it opens a multi-file review stream with sidebar navigation, split or stacked layouts, syntax highlighting, watch mode, and inline AI or agent annotations beside the code they describe. It also ships a Hunk review skill and live-session CLI so an agent can inspect the current review, navigate files and hunks, reload the view, and add comments while the human keeps the TUI open. That makes Hunk a practical fit for vibe-coding teams that now need sharper local review of agent-authored changes before they commit or open a PR.
Why Choose Hunk?
Choose Hunk if your team already uses terminal coding agents and needs a better local review surface before commits or PRs.
The inline agent annotation model is useful because review rationale sits beside the code instead of being buried in a chat transcript.
Watch mode is a good fit for agent loops where files keep changing while the human stays in review mode.
MIT licensing and simple npm, Homebrew, and Nix installs make it easy to evaluate without committing to a hosted review product.
Key Features
Review-first terminal UI for full changesets with a multi-file sidebar, split and stacked layouts, syntax highlighting, line numbers, wrapping controls, and keyboard or mouse navigation.
Agent-aware annotations that render AI reasoning beside the exact code being discussed instead of leaving review notes in a separate chat transcript.
Watch mode for keeping a Hunk session open while a coding agent edits the working tree in another terminal.
Git-style commands for reviewing working-tree diffs, commits, raw file comparisons, and patches from stdin, with support for Git, Jujutsu, and Sapling workflows.
Agent workflow support through hunk session commands and the bundled Hunk review skill, allowing an agent to inspect the live review, navigate files, reload content, and apply comments.
Distribution through npm, Homebrew, and Nix, plus a reusable OpenTUI diff component for teams building their own terminal tools.
Pros & Cons
- It addresses the practical review problem created by agent-written code instead of pretending the coding agent itself should be the only reviewer.
- The live-session workflow gives humans and agents a shared review surface, which is more useful than pasting huge raw diffs back into chat.
- MIT licensing, cross-platform support, and npm/Homebrew/Nix installs make it easy to test without buying into a proprietary review platform.
- The product signal is strong for a focused CLI: 6k+ GitHub stars, active releases, and visible use in the terminal tooling community.
- Hunk does not replace a coding agent or hosted pull-request reviewer; it is the local diff-review layer in the workflow.
- Teams still need discipline around what the agent annotates and what humans verify, because inline AI notes can be wrong just like chat output.
- The strongest fit is terminal-heavy review; developers who prefer browser or IDE review surfaces may not adopt it as their default.
- Advanced live-session agent control depends on local loopback access, which can be blocked by stricter agent sandboxes.
Detailed Use Cases for Hunk
Open Hunk on the current diff when a coding agent has changed many files and you need a structured review path instead of paging through raw patch text.
Use watch mode while an agent edits in another terminal so the review surface reloads as the working tree changes.
The bundled skill and hunk session commands let an agent inspect the review, navigate files, and apply inline comments that the human can inspect in the TUI.
Hunk can review Git-style diffs, commits, file comparisons, and stdin patches while also supporting Jujutsu and Sapling checkouts.
Who Should Use Hunk?
Developers reviewing large local diffs from Claude Code, Codex, OpenCode, or similar agents
Terminal-heavy engineers who want a better alternative to raw git diff for AI-generated changes
Teams building review gates around vibe-coded pull requests
Agent-tooling builders who want an embeddable OpenTUI diff component or live review workflow
Perfect For
Reviewing large local diffs produced by Claude Code, Codex, OpenCode, or other coding agents before committing.
Keeping a watch-mode diff open while an agent iterates so the human can inspect changes continuously.
Letting an agent leave structured inline comments in a live review session instead of scattering notes across chat.
Replacing raw git diff output with a more navigable terminal review UI for Git, Jujutsu, or Sapling workflows.
Technical Details
Hunk Comparisons & Alternatives
Hunk review
Hunk vs Delta
Hunk vs Difftastic
terminal diff viewer for AI code review
agent-authored changeset review
hunkdiff review
Developers compare Hunk with other vibe coding tools when they need a better workflow fit, not just a better landing page.
Direct Competitors
Delta
Difftastic
Open Code Review
CodeRabbit
Similar Tools You Might Like
Community Codex CLI fork that adds Auto Drive orchestration, background Auto Review, browser/CDP integration, theming, and multi-agent commands.
DoorDash open-source TUI that turns large feature requests into checkpointed multi-phase coding-agent workflows across Claude Code, Codex, and OpenCode.
Open-source CLI that gives AI coding agents visual verification through browser recordings, screenshots, logs, and PR-ready proof artifacts.
Alternative Tools to Consider
AI code review platform for pull requests, IDEs, and the CLI, built to catch bugs and keep fast-moving agentic teams from merging sloppy changes.
Open-source AI code review CLI that reads diffs, explores repo context, and produces line-level review comments for agent-heavy pull requests.
Do one more comparison before you commit to Hunk
Strong picks usually survive one more internal check. Read deeper, compare a neighbor, then leave for the vendor page if the fit still holds.