Replies: 3 comments 1 reply
-
|
Hey @v33-kind, This is incredible work. You built an interactive proposal with a visual workspace concept - that's way beyond what I usually see in feature discussions. Thank you for putting this together. I checked the demo and the phased approach makes a lot of sense. Swarm is currently in alpha and the visual workspace you're proposing aligns with where I want to take it. A few thoughts:
Your understanding of the architecture is clearly solid. Let me review the proposal in detail this weekend and come back with specific feedback on each phase and how we could go further. This is exactly the kind of input that shapes Skales - I appreciate your time and work a lot! |
Beta Was this translation helpful? Give feedback.
-
|
@v33-kind Apologies for the delay - shipped v8.0.1 with 15+ fixes and immediately had to start on v8.0.2 (a hotfix introduced a new bug) 🤦♂️ Your proposal is the most thorough piece of external input Skales has ever received. I went through the full interactive prototype again and the competitive analysis. Specific feedback on each phase: Phase 1 (Visual Workspace): Ready to start. The node/edge types you defined map directly to existing data structures. This can go into v8.2 or v9. Phase 2 (Observability): The data is already there (logs, retries, timestamps). Surfacing it is UI work, not architecture work. Realistic. Phase 3 (MCP): You're right that skills have structured I/O ready for wrapping. This is the highest ROI phase. Phase 4 (A2A): Another contributor is independently working on agent-to-agent delegation right now. Your A2A alignment would complement that work. Phase 5 (Real-Time): SSE replacing polling is overdue. Agreed. I'd like to discuss a more direct collaboration on the Swarm module. If you're interested, reach out via email: request@skales.app |
Beta Was this translation helpful? Give feedback.
-
|
Hi @talentsache Appreciate the detailed feedback on each phase - and no need to apologize for the delay. Shipping v8.0.1 with 15+ fixes while a hotfix introduces another bug is the reality of building fast. MCP as highest ROI makes sense. Good to know about the A2A contributor too - would rather complement than overlap. Skales is the first project I've ever contributed to. I've pivoted into AI product strategy and roadmap thinking - the industry I'm currently in isn't where I want to build. Getting this kind of feedback on a first contribution tells me I'm in the right place. Since posting, I went down the Discussion #34 rabbit hole and found the root cause of the agent routing issue. The task.agent field is saved but executeTask() never reads it. Built a v2 with the routing fix, 5 enhancements, and 3 bug fixes. All code commented out for review. Interactive proposal: https://v33-kind.github.io/skales/ideas/agentswarm-v2/agent-swarm-proposal-v2.html Will reach out via email, thank you. @v33-kind |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi @talentsache
Concept proposal for the Agent Swarm page - a visual workspace for local and network swarm management, with a phased roadmap covering MCP and A2A alignment.
Interactive MVP proposal viewable here:
https://v33-kind.github.io/skales/ideas/agentswarm/agent-swarm-proposal-v33kind.html
@v33-kind
kind. curious. clauding.
Beta Was this translation helpful? Give feedback.
All reactions