Building Cezar - open-source cockpit that runs Claude Code, Codex and friends 24/7 on your machine or VPS. No API keys. MIT. https://xitter.cf/t.co/U2Q5qsnYsZ
Lodz, Poland
Joined March 2024
- Tweets82
- Following99
- Followers22
- Likes62
Pinned Tweet
Poland's own Agentic Development Environment 🇵🇱
Cezar 0.16.0 is out:
🖥️ full terminal per agentic task
👀 preview mode embadded + auto-detected start commands
🎨 design mode - UI comments become agent context
30+ contributors and counting 🚀
cezar.run
Urgent question! Does anyone in your organization evaluate AI implementation based on the number of tokens consumed?
I’m shocked after the last two weeks of conversations how often companies do this!
Total mistake!
Monorepo or multi repos how you working with agents?
For me monorepo currently - not special arguments here, just it is simpler and work well.
What about you?
For me idea is pretty simple we create a well structured document with cooperation with an agent to define together how we will build functionality.
Than next agent, fresh context use it to implement as a main part of its starting context
After -> archive
How it looks for you?
Designing a product that actually helps developers avoid burnout in the age of AI?
Who's doing that?
We built Cezar for our team with our mental health in mind 😅
Three. That’s how many AI tools a human can juggle at once. Beyond that, productivity drops and serious errors rise by 39%.
Now look at your coding setup. How many agent sessions are open? How many have you lost track of?
Until now, the orchestration layer was… you. Reading logs. Answering agents’ questions. Merging PRs.
Congratulations. You automated coding and gave yourself a full-time babysitting job :)
That’s why we built Cezar, our open-source tool for managing coding agents.
Try it in our cloud with zero installation, or grab the open-source version on GitHub.
cezar.run
Patryk Lewczuk retweeted
Three. That’s how many AI tools a human can juggle at once. Beyond that, productivity drops and serious errors rise by 39%.
Now look at your coding setup. How many agent sessions are open? How many have you lost track of?
Until now, the orchestration layer was… you. Reading logs. Answering agents’ questions. Merging PRs.
Congratulations. You automated coding and gave yourself a full-time babysitting job :)
That’s why we built Cezar, our open-source tool for managing coding agents.
Try it in our cloud with zero installation, or grab the open-source version on GitHub.
cezar.run
github: github.com/open-mercato/ceza…
More about release: github.com/open-mercato/ceza…
In OSS I'm using Opus to define what to do and create concrete tasks for GPT Luna which is really good, fast and cheap.
Patryk Lewczuk retweeted
basal-omni 🇵🇱 - TOP1 🤩🤩🤩 🌍 na świecie w modelach otwartych oraz TOP2 wśród zamkniętych i otwartych
Wg niezależnego benchmarku modeli system-one gdzie testowane są wszystkjie modele na świecie - JevBench (link do bencha w pierwszym wpisie). Czyli można? Można! Polskie AI 🇵🇱
basal wyprzedził model OpenAI, Microsoft oraz pierwowzór z TypeSafe (jev).
Thank you, @airesearch12 , for the independent evaluation on JevBench.
I always had problem with writing but I'm strongly believe as well that post is an expressions of our thoughts and I want to keep them that way.
I wrote all post myself to remain truly authentic!
i love writing. i never let agents write for me. whether it's docs, or tweets, or shitposts, or even prompts (okay maybe that one's harder as agents manage other agents)
there's a simple reason for that: i still strongly believe that writing is thinking. it's a skill i hope to never give up. it's also important to me that whatever i say is always authentic, mistakes and all. my favorite posts are the ones that i just full send without any editing
i suspect it also helps with prompting. being able to clearly express what i want, and knowing the vocabulary i need to be able to direct my agents does seem to make for more effective usage of agents. in the same way that engineers have an advantage when we prompt agents to write code, because we are intimately familiar with the words that make up the discipline
knowing the right word incantations to get an agent to produce a desired goal almost feels like spellcasting. as models get more capable, and that input box looks more and more like a magic box that can output anything, words become the interface between you and the genie
there may come a time when we give up the keyboard in exchange for brain-computer interfaces. but until then, i'm writing every word.
Cezar 0.15.0 is out!
Workflows are graphs now: loops, gates, fan-out and join, and a live graph on every run, so you watch the state machine instead of scrolling logs. Your old YAML workflows still run.
Changelog: github.com/open-mercato/ceza…
Try it in any repo: npx cezar-run
I have totally different experience and I will explain why!
In my product we use custom made code review skill which has really strict rules what to check and what to report as a bug or improvement opportunity.
This way some PRs with good code can be green after one loop!
Couldn't agree more!
Same trend as with cars, for example. The more accessible something gets, the more demand it creates.
We're seeing an increase in demand for software among small and medium-sized companies, which can now afford custom-made software.
There is one thing I do differently from Matt regarding AGENTS.md.
What I add to the root AGENTS.md is a Context Task Router - a simple table indicating where to fetch additional context for various tasks.
It works really well in big projects.
Prompt of the day:
/writing-for-agents my AGENTS.md is a hot mess, propose a series of restructurings that:
- Remove no-ops
- Use progressive disclosure
- Move instructions to CODING_STANDARDS.md
Apply your work over three subagents, each more radical than the last. Create a single PR.
AGENTS.md is the most common source of token inefficiency. This prompt kills it dead.
