Refact.ai

Open-source Autonomous AI Agent that handles engineering tasks end-to-end.

⭐️ Get for IDE:

[Refact.ai @refact_ai]

Vibe Coding Series with Refact Ambassador https://t.co/dPFylOKLsa

[Refact.ai @refact_ai]

Vibe Coding Series with Refact Ambassador https://t.co/aAm7gaGvps

[Refact.ai @refact_ai]

Vibe Coding Series with Refact Ambassador https://t.co/1S5qMvo92w

[Refact.ai @refact_ai]

Weekly Showcase and Vibe Coding with Ambassador https://t.co/RcH1Qs7nGM

[Refact.ai @refact_ai]

🔥Live now: Weekly Refact Showcase!

Watch how solo dev built a memory game, complete with custom photos, fireworks, and more.

From idea to polished app, quickly — using @refact_ai Agent.

🎥 Tune in now on YouTube: https://t.co/9qhTt4e8xa

[Refact.ai @refact_ai]

Weekly Showcase with Refact Ambassador https://t.co/0xAGYET6Lk

[Refact.ai @refact_ai]

Weekly Showcase with Refact Ambassadors https://t.co/abMdk2rKqG

[Refact.ai @refact_ai]

Also live on YouTube: https://t.co/RfKykH3q68

Hit “Notify me” to not miss the session!

[Refact.ai @refact_ai]

Live on X in 2 hours: @refact_ai Weekly Showcase!

⏰Today, 5:00 PM CEST

When AI Agent handles coding, your ideas ship faster. In today's showcase:
• A smart dictionary with pronunciation
• A scanner for exposed API keys

+ Devs will walk through the tech behind each build 🙌

[Refact.ai @refact_ai]

Developers learn from experience. Their AI Agents should too.

We just published a deep dive on how long-term memory can change the way you build software with AI:

https://t.co/SJsjNASzZ4

Plus, a bonus for your team at the end 👀

[Refact.ai @refact_ai]

Ever feel like you’re stuck in Groundhog Day when using AI?

You explain your code, it helps for a while – next session, blank state, same mistakes.

🧠The solution? Memory.

It lets AI Agents:
• Keep context across sessions
• Reuse what worked
• Share knowledge team-wide

🧵

[Refact.ai @refact_ai]

Refact Weekly Showcase is live!
Plus, we have a T-shirt giveaway for the viewers 😏

Join us on YouTube, fill out the form, and get a chance to win @refact_ai merch:

https://t.co/rNmIoruNX5

[Refact.ai @refact_ai]

Weekly Showcase with Refact Ambassador https://t.co/sH5IaeW1Nk

[Refact.ai @refact_ai]

Weekly Showcase with Refact Ambassador https://t.co/sH5IaeW1Nk

[Refact.ai @refact_ai]

🔥10 mins till we go live!

[papajams @papajimjams]

Showcasing an app i am building using @refact_ai focused on breathwork & meditation in two hours

🫁 @flow_blockchain alignment obvious lol
🌳@LC #lens for social protocol features
📖 @StoryProtocol for IP registration, would love the next WIM HOF technique to be born here!

[Refact.ai @refact_ai]

We'll go live here on X at 5:00 PM CEST.

If you'd prefer to watch it on YouTube, here is the link:
https://t.co/rNmIoruNX5

Remember to hit “Notify me” so you don’t miss the event.

[Refact.ai @refact_ai]

🔥Join us tomorrow for the @refact_ai Weekly Showcase: Live on X!

⏰July 3, 5:00 PM CEST

Users will demo projects built with https://t.co/dwkVHqeI9I Agent:
• Breathing app
• Video content generator
• A full k8s recovery workflow

🎁+ Giveaway: Refact T-shirt!

[Refact.ai @refact_ai]

🧑‍💻Hack Night on Discord:
https://t.co/B28N6dngFv

Follow the link and click “Interested” to get notified when we start!

[Refact.ai @refact_ai]

We’re building live with @refact_ai Agent today 🛠️

📍July 1, 4:30 PM CEST: Hack Night on Discord.

It's a bi-weekly community coding session powered by our AI Agent.

Join to explore, suggest ideas to be implemented on the spot, or contribute to the open-source repo 👇

[Refact.ai @refact_ai]

💡pass@1 is not mandatory to be included in the SWE-bench leaderboard. But it is a realistic way to evaluate AI agents.

Here’s how @refact_ai performed:

📌SWE Verified: 74.40% — best under pass@1
📌SWE Multimodal: 35.59% — #1 overall

Tech breakdown: https://t.co/VCyyxcXFFG

[Refact.ai @refact_ai]

AI Agent that gets it right on the 1st attempt.

That’s what pass@1 means in SWE-bench and how we ran @refact_ai. One chance, no retries.

We believe this is the best way to show how reliable AI really is.

Devs need results they can trust, not AI code they have to double-check.