Now in open beta — orchestrate a team of Mel agents from your terminal. Learn more

Autonomous code review

Mel for code review

Stop waiting on the first review. Mel reviews your diff with your whole codebase as context — enforcing your repo's conventions, flagging the logic and contract issues plain linters miss, and iterating on the fix with you, right in the terminal.

Why it helps

Ship better code, faster.

Mel's review panel and agent turn the review bottleneck into a first pass that's already done by the time a human looks.

Shorter review loops

Get an immediate, actionable first pass on every diff and iterate without waiting on a teammate to free up.

Fewer escaped bugs

Mel catches convention drift, changed contracts, and logic slips that static checks and basic linters routinely miss.

Less busywork

Routine nits and obvious mistakes get surfaced — and fixed — automatically, so reviewers can focus on judgment.

How it works

From diff to inline findings, in your shell.

Mel reads the change with full-repo context, applies your standards, and keeps you in control of every edit.

  1. 01

    Open the diff

    Hit the git chip, the Code review button, or ⌃⇧= and Mel loads your uncommitted changes — or any branch base — into a clean, per-file diff.

  2. 02

    Review with codebase context

    Ask Mel to review. It greps, reads related files, and uses LSP signatures to catch contract changes and risky patterns across the repo — not just the lines that changed.

  3. 03

    Comment and iterate

    Drop an inline comment on any line and send it to the agent — it lands in the agent's prompt, ready to run. Mel then proposes the fix as an approvable diff, and you decide what lands.

Built in

Codebase context, LSP, and interactive review — built in.

Everything Mel needs to give a real review lives in the terminal you already work in.

Whole-codebase context

Mel reads beyond the diff with grep, file reads, and find — so it understands how a change ripples through the repo.

Live language intelligence

Real LSP servers (rust-analyzer, typescript-language-server, pyright, clangd) power hover signatures, go-to-definition, and inline diagnostics while you review.

Inline comments → agent

Comment on a diff line and hand it to Mel. It returns a scoped, reviewable fix instead of a vague suggestion.

Any diff base

Review uncommitted work or compare against any branch from a searchable picker — with per-file +/− and discard.

Language intelligence

Review with the language server on.

Enable the LSP for your repo and Mel reviews with real type information — hover any identifier for its signature and docs, and Ctrl+Click to jump to the definition, even into a dependency. No server installed? Mel falls back to a git-grep doc lookup.

  • rust-analyzer, typescript-language-server, pyright & clangd
  • Hover signatures + docs (300ms dwell)
  • Ctrl+Click go-to-definition
  • Same engine in the diff and the editor
mel · code-review
/agent code review
rust-analyzer, typescript-language-server, pyright & clangd
Hover signatures + docs (300ms dwell)
Ctrl+Click go-to-definition
✓ done · reviewed & approved
Code review

Stop waiting on first reviews.

Let Mel handle the first pass on every change, catch the routine issues early, and keep your reviewers focused on the calls that need a human.