Ivan Zografski retweeted
No need to change your coding agent , every time new model drops
You can get them all on same place in KiloCode
1/ kilo desktop is generally available as of today.
it's built around one idea: run more agents at once and still review every change. we've had a month with it.
1
1
49
7/ for data science: agents edit live jupyter notebooks, conda environments are built in, and you get anaconda's vetted packages. that matters when the agent is the one picking dependencies.
1
7
8/ download: kilo.ai/install
full post: blog.kilo.ai/p/kilo-desktop-…
7
Ivan Zografski retweeted
Ran out of token ?
You get free model on KiloCode, use that to continue your project
Ivan Zografski retweeted
securing agents means knowing which tools and MCP servers they can reach, and who approved that access.
blog.kilo.ai/p/anaconda-and-…
Ivan Zografski retweeted
Too many people treat AI as a zero-sum race. Open-source is winning by doing the opposite: collaboration over competition!
This is why we’re excited to release today the first part of our partnership with @OpenRouter to make open models easier to use for all! Hugging Face for storage and collaboration. OpenRouter for inference. Connected so you can:
🤗 Run OpenRouter inference directly on Hugging Face
🔓 Access Open model weights directly from OpenRouter
🤖 Store your OpenRouter agent traces on HF to inspect them, train better models or share them with the world!
More collaboration. Fewer zero-sum games. More power to AI builders. Let’s build! 💛💛💛
Ivan Zografski retweeted
At All Day AI, @madhuakula ran a coding agent on a plain laptop with no sandbox and gave it a README with a prompt injection inside. In his simulated attack, the agent read the SSH keys and AWS credentials on disk and sent them to a server he controlled.
His fix: keep the controls where the agent can't reach them, because if the agent can reach a setting, it can change it.
In Kilo you decide which commands run without asking. Keep that list short, and leave anything that touches the network on approval.
For more insights on builder perspecrtives on how to use agents securely, check: blog.kilo.ai/p/five-things-d…
Ivan Zografski retweeted
OpenAI's proposed cutoff of its models inside Cursor was set for today. As of this morning, neither company has confirmed it happened.
Thousands of developers are waiting to find out which models their editor will run, and they don't get a vote. That's the risk when your tool's model access depends on a contract you aren't part of.
In Kilo you choose the model. Bring your own key, sign in with ChatGPT to keep GPT, or switch to another model in the picker when something changes.
Ivan Zografski retweeted
Cursor Enterprise customers: get Kilo Enterprise with $0 subscription fees for up to 12 months.
We’ll match your seat count 1:1 for the rest of your eligible Cursor contract. Inference is excluded, and verification applies.
Apply by November 30, 2026: blog.kilo.ai/p/nothing-to-lo…
Ivan Zografski retweeted
november 12 is when openai stops providing models to cursor. if your team relies on GPT inside it, now’s the time to plan.
blog.kilo.ai/p/cursor-exodus
Ivan Zografski retweeted
a better architecture on paper isn’t a faster system. linear attention without engine work built for it can slow inference down. blog.kilo.ai/p/open-weights-…
Need is a good reason to reach for AI. Habit is a slippery one. Read that twice.
🔗 solveo.co/post/use-ai-becaus…








