weave-stack

name	weave-stack
description	Deploy and operate RAG infrastructure using weave-cli stacks. Use when: (1) initializing weave-stack.yaml for k8s or podman, (2) deploying VDB + ingestion + dashboard, (3) monitoring stack health, (4) day-2 operations (backup, scaling, security), (5) managing the weave dashboard UI. NOT for: initial setup (use weave-setup), data ingestion (use weave-ingest), querying (use weave-search), or evaluations (use weave-eval).
homepage	https://github.com/Maximilien-ai/weave-cli
tags	["rag","data","weave","vector-database"]
metadata	{"openclaw":{"emoji":"📦","requires":{"bins":["weave","kubectl"]},"install":[{"id":"brew","kind":"brew","formula":"Maximilien-ai/tap/weave-cli","bins":["weave"],"label":"Install weave-cli (brew)"},{"id":"go","kind":"go","package":"github.com/Maximilien-ai/weave-cli","bins":["weave"],"label":"Install weave-cli (go install)"}]}}

weave-stack

Deploy and operate complete RAG infrastructure — VDB, ingestion pipeline, dashboard, and day-2 ops.

When to Use

• Initializing stack configuration (weave-stack.yaml)
• Deploying RAG infrastructure to k8s or podman
• Monitoring stack health and logs
• Day-2 operations: backup, scaling, security patching
• Managing the weave dashboard UI

When NOT to Use

• Initial weave-cli setup -> weave-setup skill
• Data ingestion commands -> weave-ingest skill
• Search and querying -> weave-search skill
• Evaluations -> weave-eval skill

Stack Lifecycle

Initialize

weave stack init                     # Creates weave-stack.yaml
weave stack validate                 # Validate configuration

weave-stack.yaml Format

runtime:
  type: kubernetes
  provider: kind              # kind | minikube | eks | gke
  nodes: 1
  container_runtime: podman   # podman | docker

vectordb:
  type: milvus
  version: "2.3.0"
  resources:
    requests: { memory: "2Gi" }
    limits: { memory: "4Gi" }

llm:
  provider: openai
  embedding_model: text-embedding-3-small
  chat_model: gpt-4o

image_storage:
  type: minio
  endpoint: minio:9000
  bucket: weave-images

ingestion:
  workers: 2
  batch_size: 10
  retry:
    strategy: exponential
    max_retries: 3
  checkpoint: true

Deploy

# Local Kubernetes (Kind)
weave stack up --runtime kind

# Minikube
weave stack up --runtime minikube

# Cloud
weave stack up --runtime eks      # AWS EKS
weave stack up --runtime gke      # Google GKE

Monitor

weave stack status                   # Component health
weave stack logs                     # All logs
weave stack logs milvus              # Specific component
weave stack logs --follow            # Stream logs

Stack Operations

# Ingest data in deployed stack
weave stack ingest ./data/ --collection StackDocs

# Manage collections in stack
weave stack collections list
weave stack collections create NewCollection

# Backup stack data
weave stack backup

# Direct kubectl access
weave stack kubectl get pods
weave stack kubectl describe pod milvus-0

# Port forwarding
weave stack port-forward

Dashboard

weave stack dashboard                # Open web dashboard

Dashboard features: search interface, collection browser, health monitoring, agent queries.

Shutdown

weave stack down                     # Stop all services

Local Development Stacks

Docker Compose (Milvus)

# local/milvus/docker-compose.yml
services:
  milvus:
    image: milvusdb/milvus:v2.4.0-standalone
    ports:
      - "19530:19530"
      - "9091:9091"
    deploy:
      resources:
        limits:
          memory: 12G

cd local/milvus && docker compose up -d

Docker Compose (MinIO for images)

cd docker && docker compose -f docker-compose.minio.yml up -d
# API: localhost:9000, Console: localhost:9001

Day-2 Operations

Regular Updates

weave stack status                   # Check current state
weave stack backup                   # Backup before changes
# Update weave-stack.yaml version numbers
weave stack up                       # Rolling update

Scaling

# Edit weave-stack.yaml
# Increase nodes, workers, memory limits
weave stack validate
weave stack up                       # Apply changes

Security

# Rotate credentials in .env
weave config update --env
weave stack up                       # Redeploy with new creds

Backup & Recovery

weave stack backup                   # Backup all collections
weave backup validate backup.weavebak
# In case of failure:
weave backup restore backup.weavebak

Metrics & Monitoring

# Start metrics server
weave serve --metrics-port 9090

# Prometheus endpoint
curl http://localhost:9090/metrics

MCP Integration

weave mcp list                       # Available tools
weave mcp call tool-name             # Call MCP tool
weave mcp test                       # Test connection
weave mcp list --server http://localhost:8030

Decision Workflow

1. Choose runtime: kind for local dev, eks/gke for production
2. Initialize: weave stack init
3. Configure VDB, LLM, storage in weave-stack.yaml
4. Validate: weave stack validate
5. Deploy: weave stack up --runtime kind
6. Verify: weave stack status
7. Ingest data: weave stack ingest
8. Open dashboard: weave stack dashboard
9. Set up monitoring and backups for production