Deploy and orchestrate collaborative multi-agent teams using HiClaw's Manager-Workers architecture on Docker or Kubernetes with Matrix rooms for human oversight
Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Deploy and orchestrate collaborative multi-agent teams using HiClaw's Manager-Workers architecture on Docker or Kubernetes with Matrix rooms for human oversight
triggers
["set up HiClaw multi-agent collaboration platform","deploy collaborative AI agents with HiClaw","create a team of AI workers with HiClaw","configure Manager-Workers architecture for agents","install HiClaw on Kubernetes with Helm","manage AI agent teams in Matrix rooms","orchestrate multiple agents with human-in-the-loop","deploy OpenClaw and QwenPaw workers together"]
HiClaw is an open-source collaborative multi-agent runtime platform built on a Manager-Workers architecture. It enables multiple AI agents to collaborate in Matrix-based rooms with full human visibility and intervention capabilities. Workers operate with consumer tokens while real credentials stay in the centralized Higress AI gateway, providing enterprise-grade security.
The platform supports multiple agent runtimes (OpenClaw, QwenPaw/CoPaw, Hermes) working together, uses MinIO for shared file systems to reduce token consumption, and provides Kubernetes-native declarative resource management.
apiVersion:hiclaw.io/v1kind:Workermetadata:name:code-reviewernamespace:hiclaw-systemspec:runtime:openclaw# or copaw, hermesprofile:|
You are a senior code reviewer specializing in Go and Python.
Review code for security issues, performance problems, and best practices.
Provide constructive feedback with specific suggestions.
mcp:servers:-name:githubtype:githubconfig:GITHUB_PERSONAL_ACCESS_TOKEN:${GITHUB_TOKEN}-name:filesystemtype:filesystemconfig:allowedDirectories:-/workspace/codeenv:-name:REVIEW_STRICTNESSvalue:"high"
Apply the Worker:
kubectl apply -f worker-code-reviewer.yaml
Team CRD
Create a Team of Workers managed by a Team Leader:
apiVersion:hiclaw.io/v1kind:Teammetadata:name:dev-teamnamespace:hiclaw-systemspec:leader:runtime:openclawprofile:|
You are a technical lead coordinating a development team.
Break down tasks, assign to appropriate specialists, and synthesize results.
mcp:servers:-name:githubtype:githubconfig:GITHUB_PERSONAL_ACCESS_TOKEN:${GITHUB_TOKEN}members:-name:backend-devruntime:copawprofile:Gobackenddevelopmentspecialistmcp:servers:-name:filesystemtype:filesystemconfig:allowedDirectories:-/workspace/backend-name:frontend-devruntime:hermesprofile:React/TypeScriptfrontendspecialistmcp:servers:-name:filesystemtype:filesystemconfig:allowedDirectories:-/workspace/frontend
Apply the Team:
kubectl apply -f team-dev.yaml
Human CRD
Invite human users to participate:
apiVersion:hiclaw.io/v1kind:Humanmetadata:name:alicenamespace:hiclaw-systemspec:displayName:Alice(ProductManager)password:${ALICE_PASSWORD}# Set via Secret
Apply the Human:
kubectl apply -f human-alice.yaml
Working with the hiclaw CLI
HiClaw v1.1.0+ includes a CLI for managing resources:
# List all Workers
hiclaw worker list
# Create a Worker from template
hiclaw worker create --name data-analyst --runtime openclaw
# Delete a Worker
hiclaw worker delete data-analyst
# List Teams
hiclaw team list
# Create a Team
hiclaw team create --name research-team --leader-runtime copaw
# Add Worker to Team
hiclaw team add-member research-team --worker web-scraper
# View Team status
hiclaw team status research-team
# Invite human user
hiclaw human invite --username bob --display-name "Bob (Designer)"
Each runtime excels at different tasks, working together in the same room.
Upgrade
Docker Upgrade
Preserves all data (Matrix rooms, MinIO files, configuration):
# Latest version
bash <(curl -sSL https://higress.ai/hiclaw/install.sh)
# Specific version
HICLAW_VERSION=v1.1.2 bash <(curl -sSL https://higress.ai/hiclaw/install.sh)
apiVersion:hiclaw.io/v1kind:Workermetadata:name:budget-constrained-workerspec:runtime:copawprofile:Cost-consciousassistanttokenPlan:provider:qwenplan:basic# or premiummaxTokensPerRequest:4000dailyLimit:100000
Common Patterns
Pattern: Task Decomposition
Manager breaks complex task into subtasks, assigns to specialist Workers:
apiVersion:hiclaw.io/v1kind:Teammetadata:name:research-projectspec:leader:runtime:openclawprofile:|
Break down research requests into:
1. Data collection (assign to web-scraper)
2. Data analysis (assign to analyst)
3. Report generation (assign to writer)
members:-name:web-scraperruntime:hermesprofile:Webscrapinganddataextraction-name:analystruntime:copawprofile:Statisticalanalysisandinsights-name:writerruntime:copawprofile:Technicalreportwriting
kubectl port-forward -n hiclaw-system svc/minio 9000:9000
mc aliasset hiclaw http://localhost:9000 $MINIO_ROOT_USER$MINIO_ROOT_PASSWORD
mc ls hiclaw/hiclaw-shared
