YC F24's @ManicodeAI is a CLI tool that writes code for you. It uses your codebase as context and edits your files automatically. Try it out: `npm i -g manicode` ycombinator.com/launches/M2Q-m… Congrats on the launch, James Grugett and Brandon Chen!

This is incredible. We dropped a new landing page comparing coding agents. It's sooo good 😂 Codebuff versus the competition:

Hooray, it's Codebuff Launch Week! Day 1: Usage-based pricing is live - No more subscription lock-in - Buy credits $5 at a time - Credits never expire Your move, Cursor @windsurf 👀

Codebuff Launch Week — Day 2 Introducing "codebuff --max" mode, for insanely high quality results! It's an eerie hybrid of Gemini 2.5 Pro & Sonnet 3.6 that is both good at planning and taking action through tool calls. Drop it into any project and watch it build 🏗️

Grant Slatton bro just use codebuff

Codebuff Launch Week — Day 3 Introducing "codebuff --lite" mode ⚡️, for faster & lower cost results! This uses Gemini 2.5 Flash Thinking, which is currently the best bang-for-your-buck model. We're fast becoming "Gemini Code" for relying on so many Gemini models 😂

Lite mode compared to normal mode: - 1/5th the price - 4x speed And it works well for simple requests! Fun fact: As of yesterday, we were still using Claude Haiku for our --lite mode. Flash 2.5 is a major upgrade to say the least!

Codebuff 少し試したが一番賢そう！ AIを活用してコードベースを深く理解し、ターミナルから直接開発を加速できる次世代の開発支援ツール codebuff.com/referrals/ref-… Cursur, Windsurf, Cline などでVibe codingしてる人は無料お試しして欲しい。ClaudeCodeに近いのかな。それでよりコスパ良く賢い。

Codebuff Launch Week — Day 4 👏 Your coding agent should manage your dev processes (e.g. npm run dev)!👏 Shipped today: Type "init" in codebuff to initialize your project! Generates: - an overview of your project (knowledge .md) - a config file for bg processes (codebuff.json)

Codebuff Launch Week — Day 5, final day 🔥 Introducing our custom model for picking files! Codebuff picks a dozen files relevant to your prompt in the first 2 seconds. We shipped a new model for this, and it's twice as good at matching a golden dataset as before!

This is huge, as context is the most important input to a coding agent. If we can accurately pick files in two seconds, that could save the agent minutes of searching the codebase, and lots of $$$. Here you can see the total loss decreasing over training.

We beat Claude Code in our evals: 61% to 53%. Yes, evals are actually important haha And now we’re open-sourcing our entire codebase and launching an agent framework! Now you can build custom agents for your project and get better results 🔥

Codebuff Launch Day 2: Let’s build a state-of-the-art Code Reviewer agent! Crazy stuff: - a tree of 15 thinker agents - running in parallel - using 3 different models - to produce the ultimate code review! Run it now yourself: codebuff --agent codebuff/[email protected]

Codebuff Launch Day 3 Users want to know: Why did we go open source? What's our strategy? And also: What's the deal with creating our own new benchmark to show we're ahead of Claude Code 😂

Codebuff Launch Week Day 4(ish): the Codebuff SDK Now Codebuff runs in any environment, not just the CLI. Your custom agents 🤝 SDK You get power, flexibility and ease of use. We're in the age of truly personal software now. Check out this Codebuff-powered web app I built:

Today we're launching an SDK!!

Announcing our weekend competition to create and publish Codebuff agents! Help us fill out our Agent Store before we launch it on Monday! For the best agents that you all publish we'll be offering $200 of credit prizes for first, second, third place, and Community choice.

Announcing the Agent Store! It's npm for agents: - Publish specialized agents - Run them by CLI or SDK - Compose them to build higher level agents This is the end game for agents 🧠

.Relace has raised $23M to build the rails for AI code generation. This round is led by a16z, with participation from Matrix and Y Combinator. LLMs have proven they can write code—but scaling that code into production still needs better infrastructure. Relace is