Challenge
Running AI agents on real work takes more than a single LLM call: you have to dispatch tasks, supervise runs and keep cost under control. Without a cockpit, you fly blind and spending drifts.
scndcorpAI agents engineer (solo)
Deck, a multi-agent orchestration cockpit
Multi-agent orchestrator and supervision cockpit: specialized agents on a ticket board, telemetry and cost per run, n8n and MCP integration.
Impact
- Multi-agent orchestration operated daily on real work
- Telemetry and cost per run (LLM FinOps) to steer spending
- Multi-session resume with no loss, review before merge
- n8n and MCP integration
Project preview
Context
I design and operate Deck solo: a command post to run AI agents on real work, without flying blind.
What it is
A multi-agent orchestrator with a supervision cockpit:
- An orchestrator agent dispatches specialized executor agents across a ticket board.
- Every run is traced: tokens, equiv-API cost, duration.
- Multi-session resume with no loss, review before merge.
- n8n and MCP integration to plug in tools.
Stack
Next.js / TypeScript, Claude / Anthropic SDK, MCP, n8n.
Stack
The technologies used on this mission.
- Next.js (App Router)
- TypeScript
- Claude / Anthropic SDK
- MCP
- n8n
A similar project in mind?
Let's talk through your challenges and see how to turn your vision into maintainable, tested technical reality.