Mentat

AI coding assistant that works with your existing codebase from the terminal.

Open SourceSelf Hosted

0.0 (0)

About

Mentat brings AI pair programming to the terminal with a focus on coordinated, multi-file changes. Pointed at an existing repository, the assistant reads project context, then edits several files in one operation, creates new files where needed, and can run commands as part of a change, rather than returning snippets to paste by hand. It connects to hosted models such as GPT-4 and Claude and can also target local models, so teams choose their own provider. The project by AbanteAI is open source under the Apache 2.0 license and installs as a Python command line tool. It appeals to developers who want a scriptable, self-hostable coding agent that works inside their own environment rather than an IDE plugin tied to a single vendor, and it stands as one of the earlier open examples of a codebase-aware coding assistant.

Should you use Mentat?

Pick it when

Pick Mentat only if you want to study or fork an early Apache-2.0 example of a terminal assistant that makes coordinated multi-file edits, and you are comfortable maintaining a Python codebase yourself.

Look elsewhere when

Skip it for daily work: AbanteAI archived the CLI repository in January 2025 and it is no longer maintained. Aider covers the same terminal, multi-file, any-model workflow, still ships releases, and commits each change to git.

Alternatives to Mentat

Aider
Same Python terminal pair-programming approach with hosted or local models and every change committed to git, and it still ships releases, so a direct replacement.
OpenCode
Maintained MIT terminal agent with a TUI, LSP code intelligence, and many providers, a fuller agent than Mentat's multi-file edit loop.
Plandex
Targets the same multi-file changes but stages them in a reviewable diff sandbox for large projects; it needs a self-hosted server and has slowed.

Reviews (0)

Leave a Review

No reviews yet. Be the first to review!