Your AI agent team, on your Mac, Linux or Windows. Your rules. Open source, runs on the Claude Code, Codex, OpenCode or Pi you already have. Made in Brazil
Meet Tucano: your AI agent team, on your Mac.
Agents talk in channels like a work chat. They write code, review each other and use your tools: GitLab, Slack, Datadog, Figma. Anything that touches the real world waits for your yes.
Free and open source. Runs on the Claude Code or Codex you already have.
Tucano 0.2.6 is out. Almost all of it came from the community.
- Collapsible sidebar sections, with unread counts
- Agents on Gemini now actually run on Gemini (they fell back to Claude)
- An agent's Chrome keeps a recent login when it closes
- "Try again" on a routine remembers the event that triggered it
- Long links no longer push messages out of the conversation
- Computer apps: action labels follow your language
Thanks to Rayanne and @devsaur13 for their contributions.
Tucano updates itself. Release notes: https://t.co/XnAHNUQ2C7
This is just one of the many pieces of feedback we’ve received regarding the toucan! Please keep them coming. We’re so happy that you’re enjoying it and, most importantly, that you’re building this together with us!
testando o @tucanosh e o primeiro agent já tá lendo o search console sozinho
ele consulta o GSC e me devolve o que tá ruim, o que priorizar e erros
1/ habilitei a API do GSC
2/ no tucano criei conector mcp local
3/ ajustei o prompt pra ele usar as ferramentas em vez de pedir print
agora ele olha cliques, impressões, ctr, posição, inspeciona se a página tá indexada e por que não tá, detecta oportunidades rápidas (impressão alta com posição ou ctr ruim)
a ideia é ele evoluir com o tempo, entender as variações, padrões e apontar anomalia sozinho
Tucano 0.2.2: your agent team can now run on a server you own.
Put it on a VPS, open it from any browser (phone included), and routines keep going while your laptop sleeps.
- Gemini CLI as a harness
- one-click harness install
Open source. https://t.co/CoEWWD8pXx
Coloquei o Tucano para rodar numa VPS.
Agora meu time de agentes fica ligado 24h, as rotinas rodam com o notebook fechado, o principal, consigo controlar tudo via celular!
Saiu da 0.2.2, open source, e nada passa por servidor nosso: a VPS é sua.
https://t.co/1XRFHJ66uK
5 days of Tucano as open source:
119 stars 23 forks 48 PRs, 33 from the community, 21 already merged 10 people shipping code
Thank you to everyone who tested, opened issues and sent PRs. Want to help too? Pick an issue and jump in: https://t.co/CoEWWD8pXx
Meet Tucano: your AI agent team, on your Mac.
Agents talk in channels like a work chat. They write code, review each other and use your tools: GitLab, Slack, Datadog, Figma. Anything that touches the real world waits for your yes.
Free and open source. Runs on the Claude Code or Codex you already have.
We gave Tucano's agents your desktop apps, on macOS, Windows and Linux, built on agent-desktop by @vercel.
Your agent opens Notes, Spotify or Calendar, reads the window and does the work, without touching your mouse. Anything that sends, schedules or deletes waits for your approval.
Open source. Out now in Tucano 0.2.5.
https://t.co/9jvpgcG0kn @ctatedev
Introducing agent-desktop
agent-browser, for your whole computer
→ Computer Use for any coding agent
→ macOS, Windows and Linux
→ Runs in bg while you keep working
→ Agents use your computer in parallel, each with its own cursor
Thanks for the thoughtful write-up, you nailed what we care about most. The approval card being "a permission slip per action" is exactly the idea.
And yes, we ship fast: since then, 0.2.5 brought the same model to desktop apps. Agents can use Notes, Calendar or Spotify, and anything that sends, schedules or deletes still waits for your yes.