In the news
Organizing Context in a Multi-Agent Harness
LangChain · Published · 3 min read
In 30 seconds
- What happened
- LangChain introduced context modes for subagents in Deep Agents, allowing supervisors to share conversation history with worker subagents or isolate context for independent verifiers.
- Why it matters
- Matters when building multi-agent systems where subagents need to decide whether inheriting supervisor context saves work or introduces bias.
- Watch out
- Forked subagents inherit full supervisor history, which can bias independent reviewers or duplicate context unnecessarily when parallel subagents run.
- agent
- multi-agent
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.