How should a LangGraph supervisor route multiple agents with interrupts in one session?
keep__it_simple · reddit · 2026-08-18
A developer building a LangGraph app with a supervisor over Booking, Payments, Recommendations and Support agents asks for production patterns. Current design fixes routing on the first message, which breaks when users switch topics mid-session or one prompt needs multiple agents in sequence (e.g. "recommend a hotel, then book it").
Constraints: each agent has its own state and pending interrupts, no state leakage, ordered dependent tasks with parallel independent ones, permission checks per operation, new messages must not resume unrelated interrupts, and agents must return both streamed UI output and structured data. They pose 7 concrete questions: router vs supervisor vs orchestrator-worker vs subagents-as-tools; separate threadid, checkpointns or both; distinguishing new messages from interrupt responses; passing structured results between agents; per-turn task DAG vs dynamic ReAct; and whether Agent Cards/A2A help inside a single service.
More from coding & agent
- Build an AI Agent to predict deployment incidents with a complete hands-on guide — Pavan_Belagatti · 2026-08-18
- Why Agent Guardrails Fail in Multi-Agent Orchestration Chains — Dry-Presentation9814 · 2026-08-18
- LLMs retry by default and can hammer MCP servers to death — a rate-limiting walkthrough — pjausovec · 2026-08-18
- Sub8 adds a local password manager for Grok bots — secrets never touch the cloud — Daniel_Farinax · 2026-08-18
- DeepSeek open-sources new agent harness on Cordis — everything is a plugin — abhishek__AI · 2026-08-18
- pdf-brain: local PDF/Markdown knowledge base with Ollama embeddings and libSQL vector search — tom_doerr · 2026-08-18