Overview
Allow users to export a completed Agent Universe run as a self-contained JSON file that can be:
- Shared with teammates
- Replayed in Agentis to step through the agent execution visually
- Used as a debugging artifact
Export format (proposed)
{
"version": "0.2.0",
"task": "Research competitive landscape for AI coding tools",
"exportedAt": "2026-03-22T10:00:00Z",
"agents": [ { "id": "...", "name": "Researcher", "provider": "anthropic", "modelLabel": "claude-sonnet-4-6", "role": "researcher", "output": "..." } ],
"messages": [ { "fromId": "orchestrator", "toId": "agent-1", "content": "Deploy Researcher", "ts": 1234 } ],
"finalOutput": "...",
"durationMs": 42000
}
Implementation
- Add Export Run button to the post-analysis panel in
UniversePage.tsx
- Add Import & Replay button on the Universe landing view
- Replay mode: feed the JSON into state directly, animate the canvas at 2× speed stepping through messages by timestamp
Related
- Post-analysis panel already has Copy / Export MD / Export TXT — this adds Export JSON + Import
Overview
Allow users to export a completed Agent Universe run as a self-contained JSON file that can be:
Export format (proposed)
{ "version": "0.2.0", "task": "Research competitive landscape for AI coding tools", "exportedAt": "2026-03-22T10:00:00Z", "agents": [ { "id": "...", "name": "Researcher", "provider": "anthropic", "modelLabel": "claude-sonnet-4-6", "role": "researcher", "output": "..." } ], "messages": [ { "fromId": "orchestrator", "toId": "agent-1", "content": "Deploy Researcher", "ts": 1234 } ], "finalOutput": "...", "durationMs": 42000 }Implementation
UniversePage.tsxRelated