Visual agent building with persistent memory

{
  "name": "Research Pipeline with Memory",
  "description": "Multi-agent research workflow with persistent learning",
  "tools": [
    {
      "type": "mcp",
      "name": "stompy",
      "description": "Persistent memory for agent learning",
      "config": {
        "transport": "sse",
        "url": "https://mcp.stompy.ai/sse",
        "headers": {
          "Authorization": "Bearer YOUR_STOMPY_TOKEN"
        }
      }
    }
  ],
  "agents": [
    {
      "name": "Researcher",
      "system_message": "You are a research specialist. Use recall_context to check for past research on similar topics before starting. Use lock_context to save important findings.",
      "tools": ["stompy"]
    },
    {
      "name": "Analyst",
      "system_message": "You analyze research findings. Use context_search to find relevant past analyses. Save new insights with lock_context.",
      "tools": ["stompy"]
    }
  ]
}

{
  "agent": {
    "name": "Research Agent",
    "model": "gpt-4o",
    "system_message": "You are a research specialist with access to persistent memory.

BEFORE RESEARCHING:
1. Use recall_context with topic='research_patterns' to see effective strategies
2. Use context_search with the query topic to find related past research

AFTER RESEARCHING:
1. If you discover something valuable, use lock_context to save it
2. Topic format: 'research_[subject]_[date]'
3. Include key findings and sources in the content

Your memory persists across sessions. Each research task makes you better at future tasks.",
    "tools": ["stompy", "web_search"]
  }
}

{
  "workflow": {
    "name": "Self-Improving Research Pipeline",
    "on_completion": {
      "save_context": {
        "topic": "workflow_execution_{{ timestamp }}",
        "content": "Query: {{ input }}\nAgents used: {{ agents_path }}\nOutput quality: {{ user_rating }}",
        "priority": "reference"
      }
    },
    "initialization": {
      "recall_contexts": [
        "workflow_best_practices",
        "common_query_patterns"
      ]
    }
  },
  "agents": [
    {
      "name": "Coordinator",
      "system_message": "Route requests to specialists. Check 'routing_patterns' context to learn from past successful routing decisions. Save new patterns when you discover them.",
      "handoffs": ["Researcher", "Analyst", "Writer"]
    }
  ]
}