Protocol
Let subagents write durable artifacts instead of relaying everything through the coordinator
Every relay is another chance to lose detail.
When it fits
- Specialist subagents produce large outputs that are summarized through several agents before reaching the final artifact.
When to avoid it
- Shared artifact stores need access control, provenance and conflict handling when multiple agents write concurrently.
Why it matters
When a subagent creates structured work—code, report, dataset, analysis—store it directly in the shared artifact system and pass a lightweight reference to the coordinator. The coordinator should synthesize or validate the artifact without forcing the full content through multiple conversational summaries.
Steps
- Give the subagent an explicit artifact destination.
- Persist the output with a stable identity.
- Return a concise result plus artifact reference.
- Have the coordinator inspect or validate the artifact directly when needed.
An example
A research subagent writes `market_scan.json`; the lead receives the path and summary instead of a rewritten copy of every source note.
Check your result
Large specialist output can reach final production without a game of telephone through agent summaries.
Keep this limit in mind
- Shared artifact stores need access control, provenance and conflict handling when multiple agents write concurrently.
Connected ideas
Useful withUse one handoff shape for recurring transitions
Evidence and sources
Anthropic's multi-agent research architecture uses direct artifact persistence to reduce information loss and token overhead across subagent handoffs.
Shared artifact stores need access control, provenance and conflict handling when multiple agents write concurrently.
How we built our multi-agent research system · See source record