Skip to main content
Glama
orneryd

M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

by orneryd
QUICK_REFERENCE.mdβ€’4.93 kB
# Mimir + Open-WebUI Quick Reference ## πŸš€ Start/Stop ```powershell # Start everything .\start-openwebui.ps1 # Stop everything docker-compose down # View logs docker-compose logs -f # Restart just Open-WebUI docker-compose restart open-webui ``` ## 🌐 Access Points | Service | URL | Credentials | |---------|-----|-------------| | **Open-WebUI** | http://localhost:3000 | Create on first visit | | **Neo4j Browser** | http://localhost:7474 | neo4j / password | | **MCP Server Health** | http://localhost:9042/health | N/A | ## πŸ“‹ Workflow Steps ### 1. PM Creates TODO List ``` User sends request ↓ mimir-chain (PM agent) ↓ TODO list created in Neo4j ↓ Returns: todoList-xyz ``` ### 2. Parallel Execution ``` mimir-execute receives plan ↓ Agents claim tasks ↓ Execute in parallel ↓ QC verifies each ``` ### 3. Real-Time Updates ``` Pipeline polls Neo4j (1s interval) ↓ Streams task completions to UI ↓ Shows key decisions ↓ Collapsible details ``` ### 4. Final Report ``` All tasks complete ↓ Synthesize report ↓ Save to memory node ↓ Display in chat ``` ## 🎨 UI Elements ### Sidebar Format ``` 🎯 Workflow Status Progress bar (0-100%) β”œβ”€ βœ… Completed task β”œβ”€ βš™οΈ In progress task ← Active └─ ⏳ Pending task Active Agents ● WORKER (2) ● QC (1) ``` ### Chat Format ``` 🎯 PM Agent: ... [Highlighted] βœ… PM Summary: ... [Highlighted] [PM Details] β–Έ [Collapsible] ⚑ Starting Execution... [Highlighted] βœ… Task Complete: ... [Highlighted] πŸ’‘ Key Output: ... [Normal] [Task Details] β–Έ [Collapsible] πŸ“Š Final Report [Highlighted] ... report content ... πŸ’Ύ Saved: memory-xyz [Highlighted] ``` ## βš™οΈ Pipeline Configuration Edit in Open-WebUI: **Workspace β†’ Models β†’ Mimir** ```python MCP_SERVER_URL = "http://mcp-server:3000" SHOW_AGENT_CHATTER = True COLLAPSE_AGENT_DETAILS = True POLL_INTERVAL_MS = 1000 ``` ## πŸ”§ Common Tasks ### Test MCP Connection ```bash docker exec mimir-open-webui \ curl http://mcp-server:3000/health ``` ### Check Pipeline Mounted ```bash docker exec mimir-open-webui \ ls -la /app/pipelines/mimir_orchestrator.py ``` ### Query Neo4j ```bash docker exec neo4j_db \ cypher-shell -u neo4j -p password \ "MATCH (t:Todo) RETURN count(t)" ``` ### View Pipeline Logs ```bash docker logs mimir-open-webui -f --tail 100 ``` ## πŸ› Quick Fixes ### Pipeline Not Loading ```bash # Check file exists docker exec mimir-open-webui cat /app/pipelines/mimir_orchestrator.py # Restart Open-WebUI docker-compose restart open-webui ``` ### MCP Server Not Responding ```bash # Check health curl http://localhost:9042/health # Restart MCP server docker-compose restart mcp-server ``` ### Neo4j Connection Failed ```bash # Verify Neo4j is up docker logs neo4j_db --tail 50 # Test connection docker exec neo4j_db cypher-shell -u neo4j -p password "RETURN 1" ``` ## πŸ“Š File Structure ``` GRAPH-RAG-TODO/ β”œβ”€β”€ pipelines/ β”‚ β”œβ”€β”€ mimir_orchestrator.py ← Main pipeline β”‚ └── README.md ← Detailed docs β”œβ”€β”€ docker-compose.yml ← Services config β”œβ”€β”€ start-openwebui.ps1 ← Windows start β”œβ”€β”€ start-openwebui.sh ← Linux/Mac start └── ../integration/OPENWEBUI_INTEGRATION.md ← This summary ``` ## 🎯 Example Prompts ### Simple Task ``` Build a REST API with Express.js ``` ### Complex Workflow ``` Create a full-stack app with: - React frontend - Node.js backend - PostgreSQL database - JWT authentication - Docker deployment ``` ### Specific Requirements ``` Implement a microservices architecture with: - API Gateway (Kong) - User service (Node.js) - Product service (Python) - Message queue (RabbitMQ) - Monitoring (Prometheus + Grafana) ``` ## πŸ’‘ Pro Tips 1. **Watch the sidebar** - Real-time task progress 2. **Expand details** - Click collapsible sections for full context 3. **Check memory bank** - Workflows never get summarized out 4. **Use specific prompts** - More detail = better task breakdown 5. **Monitor logs** - Watch agents work in real-time ## πŸ“š Documentation Links - [Pipeline README](pipelines/README.md) - Full usage guide - [Architecture](docs/architecture/MULTI_AGENT_GRAPH_RAG.md) - System design - [Open-WebUI Docs](https://docs.openwebui.com/) - UI features - [MCP Protocol](https://spec.modelcontextprotocol.io/) - Integration spec ## πŸŽ‰ Key Benefits βœ… **No custom UI code** - Leverage 114k star project βœ… **Production ready** - Battle-tested by thousands βœ… **Real-time updates** - See agents work in parallel βœ… **Permanent storage** - Never lose workflow history βœ… **Mobile friendly** - Works on all devices βœ… **Zero configuration** - Works out of the box --- **Quick Start**: `.\start-openwebui.ps1` β†’ Open http://localhost:3000 β†’ Start chatting! πŸš€

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/orneryd/Mimir'

If you have feedback or need assistance with the MCP directory API, please join our Discord server