PearAI Review: The Open-Source AI Editor Fork, One Year On

PearAI shipped its first public build in mid-2024 as an open-source, AI-native fork of VS Code, rode a wave of Y Combinator (S24) attention, and within weeks became the most argued-about editor on Hacker News — for the wrong reasons. A year later it is still shipping, still open-source, and still living down its debut. We spent time in the current build to answer one question: is PearAI a real alternative to Cursor and Continue, or a marketing layer on top of code other people wrote?

What PearAI actually is

Strip the branding and PearAI is Code - OSS (the open-source core of VS Code) with a curated bundle of AI features layered on top. You get the familiar VS Code window, settings, and keybindings, plus:

• A chat sidebar for asking questions about your code, with bring-your-own-key support for Anthropic, OpenAI, and local models.
• Inline edits triggered from the editor — the same Cmd+K pattern Cursor popularized.
• An agentic “Creator” mode descended from the aider / Roo Code lineage, which plans and applies multi-file changes.
• A Perplexity-style search panel for pulling answers and docs without leaving the editor.
• Autocomplete wired through Supermaven, and project memory via mem0.

The pitch is curation. Instead of installing Continue, wiring up aider in a terminal, bolting on an autocomplete provider, and juggling API keys across all of them, PearAI assembles a working stack out of the box. That is a genuinely useful thing — if you trust the assembler.

The controversy that shaped year one

PearAI’s first month is the reason most developers already have an opinion about it. Two things happened.

First, critics noticed the early build was, functionally, VS Code plus a re-skinned Continue extension. The “AI editor” framing collided with the reality that most of the AI came from an existing open-source project. For a launch that leaned on novelty, “fork of a fork” stuck as the headline.

Second, and more damaging, was a licensing misstep. The project initially shipped forked open-source code under a custom, restrictive license rather than honoring the permissive terms it inherited. The backlash was immediate, the license was reverted to a standard open one, and the founders apologized — but the episode landed hard precisely because openness is the entire value proposition. An “open-source editor” that fumbles a license is mishandling the one thing it is selling.

Who it’s for, one year on

The honest read after a year is that PearAI is an aggregator, and you should judge it as one.

Its strengths are real. The bring-your-own-key model means you are not locked to a single vendor’s pricing or model roster — you can point chat at Claude, send agent work to a cheaper model, and pay providers directly. The bundling saves the afternoon you would otherwise spend assembling Continue, aider, and an autocomplete provider yourself. And because it is Apache-licensed VS Code underneath, your muscle memory transfers intact.

Its weaknesses are also structural. An aggregator inherits the bugs and release cadence of everything it aggregates, so polish lags a vertically integrated product like Cursor, where one team owns the whole loop. If your priority is the smoothest possible tab-complete-and-agent experience and you do not mind a subscription, Cursor is still the more finished tool. PearAI’s case is openness, model freedom, and not handing your codebase to a single closed vendor.

So: if you already pay for Cursor and it works, PearAI will not pull you away on features alone. If you assemble your own open-source AI stack, or you want model and vendor freedom on a familiar VS Code base, PearAI in its second year is worth a serious trial — with the license history as a standing reminder to read the fine print.