Agents Shouldn't Pass Context. They Should Share Memory

说明与用途

多代理链经常导致上下文膨胀，增加成本和延迟。解决方案是使用共享内存架构，包含三个层次（情节、语义、程序）并采用基于令牌预算的检索，保持提示大小固定，成本可预测。

问题

Multi‑agent chains accumulate token‑heavy context, driving up cost, latency, and diluting attention.

结果

Token consumption per round became a fixed budget, eliminating linear growth; cost and latency stabilized across long chains and multi‑session use.

工作流

• 1. Agent writes output to an append‑only episodic log (L1).
• 2. Asynchronously project new log entries into the semantic tier (L2) via LLM extraction.
• 3. Agent retrieves a token‑budgeted slice from both L1 (fresh events) and L2 (distilled facts).
• 4. Agent performs its task and appends its result to L1.
• 5. Repeat for each agent in the chain; on new sessions, L2 provides cross‑session beliefs.
• 6. Periodically rebuild L2/L3 from L1 if needed for consistency.

证据

Agents do not hand anything to each other directly. Each agent writes its output to a shared memory layer, and the next agent retrieves only the slice it needs, capped to a token budget.

来源

The episodic tier (L1) is an append‑only log of everything that happens in a session…

来源

If Agent A produced forty thousand tokens of investigation and Agent B needs the three thousand tokens of conclusion, B receives three thousand.

来源