Release 1.3 will be out in a few minutes 🎉

🚀 Hey Community, we’ve just shipped a major release: Archestra 1.3.0, which we’re calling Lyra.

The full release is available now, and over the next few days we’ll highlight one of its new features each day.

We’re starting with Skills and Sandboxes.

Agents can now load SKILL.md-based Skills and run their code in secure Sandboxes governed by the environment and access policies you choose.

Give it a try - we’d love to hear what you build with them!

hey Team!👋

I'm planning to submit a PR with this commit (github.com/archestra-ai/archestra/commit/4b153a675484…) to support tools that run longer than the standard 60-second timeout. I aimed for a minimal implementation.

However, judging by the issues in modelcontextprotocol/typescript-sdk, using resetTimeoutOnProgress=true with an onprogress callback seems like a cleaner way to handle this.

Would you recommend sticking with my minimal hotfix for now, or should we implement proper configuration parameters for the MCP client from the start?