### Example Skill Search Result and Install Command

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/hermes-worker-agent/skills/find-skills/SKILL.md

Example output from a skill search, showing the skill name and its description, along with the command to install it.

```bash
Install with ${HERMES_HOME}/skills/find-skills/scripts/hiclaw-find-skill.sh install <skill>

vercel-react-best-practices
└ React and Next.js performance guidance
```

--------------------------------

### Example: Setup MCP Proxy with No Auth

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/mcp-server-management/references/setup-mcp-proxy.md

This example demonstrates how to set up an MCP proxy server without any authentication headers.

```bash
bash .../setup-mcp-proxy.sh sentry https://mcp.sentry.dev/mcp http
```

--------------------------------

### Example: Setup MCP Proxy with API Key Auth

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/mcp-server-management/references/setup-mcp-proxy.md

This example shows how to configure an MCP proxy server using an API key for authentication.

```bash
bash .../setup-mcp-proxy.sh asana https://mcp.asana.com/sse sse \
    --header "X-API-Key: my-key"
```

--------------------------------

### Example: Setup MCP Proxy with Multiple Headers

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/mcp-server-management/references/setup-mcp-proxy.md

Configure an MCP proxy server with multiple custom headers, such as authorization and tenant ID.

```bash
bash .../setup-mcp-proxy.sh custom https://mcp.example.com/mcp http \
    --header "Authorization: Bearer xxx" \
    --header "X-Tenant-Id: my-tenant"
```

--------------------------------

### Install Cron Package and Start Crond

Source: https://github.com/agentscope-ai/hiclaw/blob/main/changelog/v1.0.4.md

Installs the cron package and starts crond in Manager and Worker.

```git-commit
feat(openclaw-base): install cron package in base image, start crond in Manager (supervisord) and Worker (entrypoint)
```

--------------------------------

### Search for Skills Example

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/find-skills/SKILL.md

Use the find command with a relevant query to search for skills. The command returns results with installation instructions.

```bash
"${FIND_SKILLS_SCRIPT}" find react performance
```

```bash
"${FIND_SKILLS_SCRIPT}" find pr review
```

```bash
"${FIND_SKILLS_SCRIPT}" find changelog
```

```bash
Install with /root/.copaw-worker/${HICLAW_WORKER_NAME}/.copaw/active_skills/find-skills/scripts/hiclaw-find-skill.sh install <skill>

vercel-react-best-practices
└ React and Next.js performance guidance
```

--------------------------------

### Present Skill Options to User

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/find-skills/SKILL.md

Example response format for presenting found skills to the user, including name, function, installation command, and a link for more information.

```bash
I found a skill that might help! The "remotion-best-practices" skill provides
best practices for Remotion video creation in React.

To install it:
"${FIND_SKILLS_SCRIPT}" install remotion-best-practices
```

--------------------------------

### Quick Start Installation (macOS/Linux)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/install/README.md

Use this command to quickly install HiClaw on macOS or Linux systems. Ensure bash and curl are available.

```bash
bash <(curl -fsSL https://raw.githubusercontent.com/agentscope-ai/HiClaw/main/install/hiclaw-install.sh)
```

--------------------------------

### Example: Setup MCP Proxy with Bearer Token Auth

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/mcp-server-management/references/setup-mcp-proxy.md

Use this command to set up an MCP proxy server that requires a Bearer token for authentication.

```bash
bash .../setup-mcp-proxy.sh notion https://mcp.notion.com/mcp http \
    --header "Authorization: Bearer ntn_xxx"
```

--------------------------------

### Install a Skill

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/find-skills/SKILL.md

Use this command to install a skill. The default installation location is `~/.agents/skills/`.

```bash
"${FIND_SKILLS_SCRIPT}" install <skill>
```

--------------------------------

### Batch Deployment Example

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/declarative-resource-management.md

Example YAML configuration for batch deployment of resources like Teams, Workers, and Humans.

```APIDOC
## Batch Deployment

Use `---` separators to define multiple resources in one file. **`hiclaw apply -f` applies documents sequentially in the order they appear** — it does not sort by kind. Put Teams before Humans that list `accessibleTeams`, and create standalone Workers before Humans that list `accessibleWorkers`.

Deletion order is not automatic: use `hiclaw delete` per resource (respect dependencies: e.g. delete Humans before Teams they reference, if your deployment requires it).

```yaml
# company-setup.yaml

# --- Team definitions ---
apiversion: hiclaw.io/v1beta1
kind: Team
metadata:
  name: product-team
spec:
  description: Product development team
  leader:
    name: product-lead
    model: claude-sonnet-4-6
  workers:
    - name: backend-dev
      model: claude-sonnet-4-6
      skills: [github-operations, git-delegation]
      mcpServers:
        - name: github
          url: https://gateway.example.com/mcp-servers/github/mcp
    - name: frontend-dev
      model: claude-sonnet-4-6
      skills: [github-operations]
    - name: qa-engineer
      model: claude-sonnet-4-6
---
apiversion: hiclaw.io/v1beta1
kind: Team
metadata:
  name: ops-team
spec:
  description: Operations team
  leader:
    name: ops-lead
    model: claude-sonnet-4-6
  workers:
    - name: monitor
      model: claude-sonnet-4-6
---
# --- Standalone Worker ---
apiversion: hiclaw.io/v1beta1
kind: Worker
metadata:
  name: admin-assistant
spec:
  model: claude-sonnet-4-6
---
# --- Human users ---
apiversion: hiclaw.io/v1beta1
kind: Human
metadata:
  name: zhang-san
spec:
  displayName: Zhang San
  email: zhangsan@example.com
  permissionLevel: 2
  accessibleTeams: [product-team]
  note: Product manager
---
apiversion: hiclaw.io/v1beta1
kind: Human
metadata:
  name: li-si
spec:
  displayName: Li Si
  email: lisi@example.com
  permissionLevel: 2
  accessibleTeams: [product-team]
  note: Backend developer
---
apiversion: hiclaw.io/v1beta1
kind: Human
metadata:
  name: wang-wu
spec:
  displayName: Wang Wu
  email: wangwu@example.com
  permissionLevel: 3
  accessibleWorkers: [admin-assistant]
  note: Administrative staff
```

One-command deployment:

```bash
bash install/hiclaw-apply.sh -f company-setup.yaml
```

For subsequent changes, edit the YAML and re-apply. To remove a resource, use `hiclaw delete <kind> <name>` (or the REST API).
```

--------------------------------

### Example: Switch with Custom Context Window

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/model-switch/SKILL.md

Example of switching the Manager Agent's model to 'my-custom-model' with a specified context window of 300,000 tokens.

```bash
bash /opt/hiclaw/agent/skills/model-switch/scripts/update-manager-model.sh my-custom-model --context-window 300000
```

--------------------------------

### File Sync Example Workflow

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/file-sync/SKILL.md

This example demonstrates a typical workflow involving task assignment, file synchronization, work completion, and result pushing.

```bash
# Coordinator assigns task: "New task [st-01]. Please file-sync and read shared/tasks/st-01/spec.md"
# Run file-sync to pull latest
copaw-sync

# Read the spec
cat shared/tasks/st-01/spec.md

# ... do the work ...

# Push results
bash ./skills/file-sync/scripts/push-shared.sh tasks/st-01/ --exclude "spec.md" --exclude "base/"

# Confirm to coordinator
"Task complete. Results pushed to MinIO."
```

--------------------------------

### Example File Sync Workflow

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/hermes-worker-agent/skills/file-sync/SKILL.md

This example demonstrates a typical workflow involving task assignment, file synchronization, reading task specifications, performing work, and pushing results back to MinIO.

```bash
# Coordinator assigns task: "New task [st-01]. Please file-sync and read shared/tasks/st-01/spec.md"
# Run file-sync to pull latest
hermes-sync

# Read the spec
cat shared/tasks/st-01/spec.md

# ... do the work ...

# Push results
bash ./skills/file-sync/scripts/push-shared.sh tasks/st-01/ --exclude "spec.md" --exclude "base/"

# Confirm to coordinator
"Task complete. Results pushed to MinIO."
```

--------------------------------

### Execution Order Example

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/team-leader-agent/skills/team-project-management/references/plan-format.md

Details the sequential execution waves for the example web app project's DAG, based on task dependencies.

```text
1. **Wave 1** (parallel): st-01, st-02
2. **Wave 2** (after st-01+st-02): st-03; (after st-02): st-04
3. **Wave 3** (after st-03): st-05; (after st-03+st-04): st-06

```

--------------------------------

### Example Response When No Skills Are Found

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/find-skills/SKILL.md

This example shows how to inform the user that no skills were found and offer direct assistance or suggest creating a new skill.

```text
I searched for skills related to "xyz" but didn't find any matches.
I can still help you with this task directly! Would you like me to proceed?

If this is something you do often, you could create your own skill:
skills init my-xyz-skill
```

--------------------------------

### Direct Skill Installation

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/find-skills/SKILL.md

Install a specific skill directly if the user provides a concrete name and requests import from the market. Always use the exact command from search results.

```bash
"${FIND_SKILLS_SCRIPT}" install <skill>
```

```bash
"${FIND_SKILLS_SCRIPT}" install remotion-best-practices
```

```bash
"${FIND_SKILLS_SCRIPT}" install github-operations
```

--------------------------------

### Skill Script Migration Example

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/design/hiclaw-controller-refactor.md

An example demonstrating the old method of managing worker creation via shell scripts versus the new approach of calling the Hiclaw CLI.

```bash
# 旧方式：create-worker.sh 内部直接操作 Matrix/Higress/Docker
bash ./skills/worker-management/scripts/create-worker.sh --name alice --model qwen3.5-plus
```

--------------------------------

### Example: Switch with Reasoning Disabled

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/model-switch/SKILL.md

Example of switching the Manager Agent's model to 'deepseek-chat' with reasoning disabled.

```bash
bash /opt/hiclaw/agent/skills/model-switch/scripts/update-manager-model.sh deepseek-chat --no-reasoning
```

--------------------------------

### Run Manager Agent Installation Script

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/README.md

Execute the installation script to run the manager agent. This is the preferred method over direct execution.

```bash
../install/hiclaw-install.sh manager
```

--------------------------------

### Example Tool Documentation within Skill

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/worker-agent/skills/mcporter/SKILL.md

Document a specific tool within a skill's SKILL.md file, including its purpose, an example call, and a table detailing its parameters.

```markdown
# Weather Operations

## Overview

Use this skill to query weather data via the centralized MCP Server.

## get_weather

Get current weather for a city:

\```bash
mcporter call mcp-weather.get_weather city=Tokyo units=metric
\```

| Parameter | Type | Required | Description |
|-----------|------|----------|-------------|
| city | string | yes | City name |
| units | string | no | Temperature units (default: metric) |
```

--------------------------------

### Install Migration Skill on OpenClaw

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/import-worker.md

Copy the migration skill to your OpenClaw's skills directory or use the install command.

```bash
cp -r migrate/skill/ ~/.openclaw/workspace/skills/hiclaw-migrate/
```

```bash
Install the hiclaw-migrate skill from /path/to/hiclaw/migrate/skill/
```

--------------------------------

### Install HiClaw with Qwen Provider

Source: https://github.com/agentscope-ai/hiclaw/blob/main/README.md

Install HiClaw specifically for Qwen (通义千问) by setting the `llmProvider` to `qwen` and specifying the appropriate API key and model.

```bash
helm install hiclaw higress.io/hiclaw \
  -n hiclaw-system --create-namespace \
  --render-subchart-notes \
  --set credentials.llmApiKey=<your-qwen-api-key> \
  --set credentials.llmProvider=qwen \
  --set credentials.defaultModel=qwen3.5-plus \
  --set credentials.adminPassword=<your-admin-password> \
  --set gateway.publicURL=http://localhost:18080
```

--------------------------------

### Example Web App Project DAG Task Plan

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/team-leader-agent/skills/team-project-management/references/plan-format.md

An example illustrating the DAG Task Plan format for a web application project, including task titles, assignments, and dependencies.

```markdown
## DAG Task Plan

- [ ] st-01 — Design database schema (assigned: @alice:domain)
- [ ] st-02 — Design API specification (assigned: @alice:domain)
- [ ] st-03 — Implement backend API (assigned: @alice:domain, depends: st-01, st-02)
- [ ] st-04 — Implement frontend pages (assigned: @bob:domain, depends: st-02)
- [ ] st-05 — Write unit tests (assigned: @charlie:domain, depends: st-03)
- [ ] st-06 — Integration testing (assigned: @charlie:domain, depends: st-03, st-04)

```

--------------------------------

### Manual Docker and Podman Agent Start

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/development.md

Use these commands to manually start the Hiclaw manager agent with either Docker or Podman, ensuring the correct socket path and environment variables are set.

```bash
# Docker
docker run -d --name hiclaw-manager \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -e HICLAW_WORKER_IMAGE=hiclaw/worker-agent:latest \
  ... \
  hiclaw/manager-agent:latest
```

```bash
# Podman
podman run -d --name hiclaw-manager \
  -v /run/podman/podman.sock:/var/run/docker.sock \
  --security-opt label=disable \
  -e HICLAW_WORKER_IMAGE=hiclaw/worker-agent:latest \
  ... \
  hiclaw/manager-agent:latest
```

--------------------------------

### Example mcporter call within skill documentation

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/mcporter/SKILL.md

This is an example of how to document a specific tool's usage within a `SKILL.md` file. It shows the `mcporter call` command and its parameters.

```bash
mcporter call mcp-weather.get_weather city=Tokyo units=metric
```

--------------------------------

### Example: Setting up a Weather API MCP Server

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/mcp-server-management/references/custom-yaml-guide.md

Demonstrates how to generate a YAML configuration file for a weather API and deploy it using the `setup-mcp-server.sh` script.

```bash
cat > /tmp/mcp-weather.yaml << 'YAML'
server:
  name: weather-mcp-server
  config:
    accessToken: ""
tools:
- name: get_weather
  description: "Get current weather for a city"
  args:
  - name: city
    description: "City name (e.g., London, Tokyo)"
    type: string
    required: true
  - name: units
    description: "Temperature units"
    type: string
    required: false
    default: "metric"
  requestTemplate:
    url: "https://api.openweather.com/v1/weather?q={{.args.city}}&units={{.args.units}}"
    method: GET
    headers:
    - key: X-API-Key
      value: "{{.config.accessToken}}"
YAML

bash .../setup-mcp-server.sh weather "<key>" --yaml-file /tmp/mcp-weather.yaml

```

--------------------------------

### Install HiClaw Manager (Windows)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/install/README.md

Execute this PowerShell script to install the HiClaw Manager agent on Windows. It will guide you through the setup.

```powershell
.\hiclaw-install.ps1 manager
# or simply
.\hiclaw-install.ps1
```

--------------------------------

### Get Remote Worker Install Command

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/worker-management/references/lifecycle.md

Retrieves the command needed to install a Worker on a remote machine. Provide the output verbatim.

```bash
bash /opt/hiclaw/agent/skills/worker-management/scripts/get-worker-install-cmd.sh --worker <name>
```

--------------------------------

### Install HiClaw on macOS/Linux

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-1.0.4-release.md

Use this bash command to install or upgrade HiClaw on macOS and Linux systems. The script will guide you through the process interactively.

```bash
bash <(curl -sSL https://higress.ai/hiclaw/install.sh)

```

--------------------------------

### Install HiClaw on macOS/Linux

Source: https://context7.com/agentscope-ai/hiclaw/llms.txt

Use this command for an interactive setup on macOS or Linux. For non-interactive installs in development or CI, set environment variables for LLM credentials and admin details before running 'make install'.

```bash
# macOS / Linux — interactive installer
bash <(curl -sSL https://higress.ai/hiclaw/install.sh)
```

```bash
# Minimal non-interactive install (developers, CI)
HICLAW_LLM_API_KEY="sk-your-key" \
HICLAW_LLM_PROVIDER="openai-compat" \
HICLAW_DEFAULT_MODEL="gpt-4o" \
HICLAW_ADMIN_USER="admin" \
HICLAW_ADMIN_PASSWORD="yourpassword" \
make install
```

--------------------------------

### Set up MinIO Alias

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/development.md

Sets up an alias for the MinIO object storage service. Replace `<user>` and `<password>` with your MinIO credentials.

```bash
mc alias set test http://localhost:9000 <user> <password>
```

--------------------------------

### Example HiClaw Manager Output

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-announcement.md

This output is displayed after a successful HiClaw installation, providing the URL to access the Manager UI and login credentials.

```text
=== HiClaw Manager Started! ===

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
  ★ Open the following URL in your browser to start:                           ★
                                                                                
    http://127.0.0.1:18088/#/login
                                                                                
  Login with:                                                                   
    Username: admin
    Password: [auto-generated password]
                                                                                
  After login, start chatting with the Manager!
    Tell it: "Create a Worker named alice for frontend dev"
    The Manager will handle everything automatically.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
```

--------------------------------

### Simulate SaaS Product Development with AI Workers

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-announcement.md

This example simulates the entire lifecycle of building a SaaS product using AI workers. It shows how to create workers, initiate a project, assign tasks, manage dependencies, handle revisions, and monitor progress from development to launch and analytics.

```text
You: Create 4 Workers:
     - alex: product manager
     - sam: full-stack developer
     - taylor: content & marketing
     - jordan: data analyst

Manager: Done. 4 Workers created, each with independent skills and memory.

You: Kick off project: AI Writing Assistant MVP, launch target this week

Manager: Got it, planning now...

         [Creates project room, invites all Workers]
         [Drafts project plan]
         
         Here's the plan, please confirm:
         
         Phase 1: Product Definition
         - task-001: Competitive research & PRD (alex)
         
         Phase 2: Development
         - task-002: Tech stack & architecture (sam, depends on task-001)
         - task-003: Core feature development (sam, depends on task-002)
         - task-004: Product review (alex, depends on task-003)
         
         Phase 3: Launch Prep
         - task-005: Launch assets (taylor, depends on task-001)
         
         Confirm to start assigning.

You: Confirmed

Manager: [Sets project status to active]
         
         @alex New task [task-001]: Competitive research & PRD
         Goal: Output PRD defining MVP feature scope
         Spec: ~/hiclaw-fs/shared/tasks/task-001/spec.md
         Report back in the project room when done

[2 hours later]

alex: @manager task-001 complete
      PRD done: core features are AI autocomplete, multi-model switching, history
      Result: ~/hiclaw-fs/shared/tasks/task-001/result.md

Manager: [Reads result.md, status: SUCCESS]
         [Updates plan, marks task-001 complete]
         
         @sam New task [task-002]: Tech stack & architecture
         PRD reference: ~/hiclaw-fs/shared/tasks/task-001/result.md
         
         @taylor New task [task-005]: Launch assets
         Product positioning: ~/hiclaw-fs/shared/tasks/task-001/result.md

[4 hours later]

sam: @manager task-002 complete
     Stack: Next.js + Vercel + Supabase
     Estimated 8 hours for core development

Manager: [Updates plan, assigns task-003]
         
         @sam Continue with core feature development [task-003]

[8 hours later]

sam: @manager task-003 complete, deployed to Vercel
     Demo: https://xxx.vercel.app

Manager: [Updates plan, assigns task-004]
         
         @alex Please review [task-004]: Product acceptance
         Test URL: https://xxx.vercel.app
         Verify feature completeness against PRD

[Issue found during review]

alex: @manager task-004 needs revision
      Issue: multi-model switching has no onboarding — users don't know how to pick
      Suggestion: add a model comparison page
      Result: ~/hiclaw-fs/shared/tasks/task-004/result.md
      Status: REVISION_NEEDED → back to task-003

Manager: [Reads result.md, status: REVISION_NEEDED]
         [Creates revision task task-006]
         
         @sam task-004 found issues, revision needed
         
         Revision task [task-006]: Add model selection onboarding page
         Feedback: ~/hiclaw-fs/shared/tasks/task-004/result.md
         Report back when done

[Revision complete]

sam: @manager task-006 complete, onboarding page added

Manager: [task-006 done, task-004 re-review passed]
         [All development tasks complete]
         
         @you MVP development complete!
         - Deployed: https://xxx.vercel.app
         - Product review passed
         - Launch assets ready
         
         Project plan: ~/hiclaw-fs/shared/projects/proj-xxx/plan.md

You: Ready to launch. @taylor Product Hunt tomorrow

taylor: Got it, scheduled. Assets are ready.

[Launch day — Manager auto-monitors progress]

Manager: @you Product Hunt launch update
         Current rank: #3
         Upvotes: 423
         Comments: 87
         
         @jordan please set up analytics tracking

jordan: On it, configuring GA4 + custom events...

[Data ready]

jordan: @manager tracking setup complete
        Dashboard: https://analytics.google.com/xxx
        
        Day 1 data:
        - Signups: 1,247
        - Day 2 retention: 34%
        - AI autocomplete usage: 78%
        - Multi-model switching usage: 23%

Manager: @you Project "AI Writing Assistant MVP" — Day 1 Report
         
         Key metrics:
         - Day 1 signups: 1,247
         - Day 2 retention: 34%
         - Feature usage: autocomplete 78%, model switching 23%
         
         Insight: model switching usage is low
         Suggestion: @alex investigate and optimize the onboarding flow

[And so it goes — Manager runs the whole loop: plan → assign → monitor → coordinate → report]
```

--------------------------------

### Quick Start Command for Manager

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/windows-deploy.md

Use this command in the Manager chat to initiate the creation of your first Worker, named 'alice'.

```text
Create a frontend developer Worker named alice
```

--------------------------------

### Worker Notification Message

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/team-leader-agent/skills/team-project-management/references/task-lifecycle.md

This is an example message to be sent to a worker via a team room. It includes task details, a link to the specification, and instructions for starting the work.

```text
@alice:{domain} New task [st-01]: Design database schema
Pull spec: shared/tasks/st-01/spec.md
Please file-sync, read the spec, create plan.md before starting. @mention me when complete.
```

--------------------------------

### Risk: Embedded Mode Controller Container Splitting

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/design/hiclaw-controller-refactor.md

Mitigation for increased deployment complexity due to splitting the controller container in embedded mode, using Docker Compose for orchestration and an install script for automated setup.

```markdown
Docker Compose 编排，install 脚本自动处理容器间网络和依赖
```

--------------------------------

### Install and Uninstall HiClaw Components

Source: https://github.com/agentscope-ai/hiclaw/blob/main/tests/skills/hiclaw-test/SKILL.md

Perform individual installation or uninstallation of HiClaw. Use `HICLAW_YOLO=1 make install` for installation and `make uninstall` for uninstallation. Existing installations can be used to run tests by skipping the build step.

```bash
# Install only
set -a && . ~/hiclaw-manager.env && set +a && HICLAW_YOLO=1 make install

# Uninstall only
make uninstall

# Run tests using existing installation (skip reinstall)
set -a && . ~/hiclaw-manager.env && set +a
./tests/run-all-tests.sh --skip-build --use-existing
```

--------------------------------

### Load Config and Run Full Tests

Source: https://github.com/agentscope-ai/hiclaw/blob/main/tests/skills/hiclaw-test/SKILL.md

Load environment variables from the configuration file and execute the full test suite, which includes installation, testing, and uninstallation.

```bash
# Load config and run tests (automatically executes install → test → uninstall)
set -a && . ~/hiclaw-manager.env && set +a && make test
```

--------------------------------

### Install HiClaw on Windows

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-announcement.md

This PowerShell script installs HiClaw on Windows. Ensure you have PowerShell 7+ and Docker Desktop installed prior to execution.

```powershell
Set-ExecutionPolicy Bypass -Scope Process -Force; $wc=New-Object Net.WebClient; $wc.Encoding=[Text.Encoding]::UTF8; iex $wc.DownloadString('https://higress.ai/hiclaw/install.ps1')
```

--------------------------------

### Install copaw-worker

Source: https://github.com/agentscope-ai/hiclaw/blob/main/copaw/README.md

Install the copaw-worker package using pip.

```bash
pip install copaw-worker
```

--------------------------------

### Install HiClaw on Windows (PowerShell)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-1.0.4-release.md

Execute this PowerShell command to install or upgrade HiClaw on Windows with PowerShell 7+. It sets the execution policy and downloads the installation script.

```powershell
Set-ExecutionPolicy Bypass -Scope Process -Force; $wc=New-Object Net.WebClient; $wc.Encoding=[Text.Encoding]::UTF8; iex $wc.DownloadString('https://higress.ai/hiclaw/install.ps1')

```

--------------------------------

### Troubleshoot Hiclaw Installation on Windows

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/faq.md

If the PowerShell installation script closes immediately on Windows, ensure Docker Desktop is installed and running. The script requires a loaded Docker daemon to connect to.

--------------------------------

### Example: Switch to Claude Sonnet

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/model-switch/SKILL.md

Example of switching the Manager Agent's model to 'claude-sonnet-4-6'.

```bash
bash /opt/hiclaw/agent/skills/model-switch/scripts/update-manager-model.sh claude-sonnet-4-6
```

--------------------------------

### Install CoPaw Worker Runtime

Source: https://github.com/agentscope-ai/hiclaw/blob/main/changelog/v1.0.4.md

Installs the CoPaw Worker runtime package via pip for remote deployment.

```bash
--runtime copaw --remote
```

--------------------------------

### Get Ready Tasks and Initialize Task Directory

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/team-leader-agent/skills/team-project-management/references/dag-execution.md

Fetches tasks that are ready for execution based on the DAG plan and prepares the necessary directory structure for a new task.

```bash
# Get ready tasks
READY=$(bash ./skills/team-project-management/scripts/resolve-dag.sh \
  --plan /root/hiclaw-fs/shared/projects/{project-id}/plan.md \
  --action ready)

# For each ready task, create task files
TASK_ID="st-01"
TASK_DIR="/root/hiclaw-fs/shared/tasks/${TASK_ID}"
mkdir -p "${TASK_DIR}"
```

--------------------------------

### Example Git Request for Cloning and Branching

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/worker-skills/git-delegation/SKILL.md

An example of a git-request message to clone a repository and create a new branch.

```bash
@manager:DOMAIN task-20260225 git-request:
workspace: /root/hiclaw-fs/shared/tasks/task-20260225/workspace
operations:
  - git clone https://github.com/higress-group/hiclaw.git
  - cd hiclaw && git checkout -b feature-xyz
---CONTEXT---
Starting work on feature XYZ
---END---
```

--------------------------------

### Deploy and Verify Batch Configuration

Source: https://context7.com/agentscope-ai/hiclaw/llms.txt

Applies the comprehensive company setup and verifies the deployment of workers and teams. The verification commands provide a summary of deployed resources and their states.

```bash
# Deploy everything
bash install/hiclaw-apply.sh -f company-setup.yaml

# Verify
docker exec hiclaw-manager hiclaw get workers
# NAME           PHASE    MODEL              TEAM           RUNTIME
# backend-dev    Running  claude-sonnet-4-6  product-team   openclaw
# frontend-dev   Running  claude-sonnet-4-6  product-team   openclaw
# qa-engineer    Running  claude-sonnet-4-6  product-team   hermes
# admin-assist.  Running  qwen3.5-plus       -              openclaw

docker exec hiclaw-manager hiclaw get teams
# NAME          PHASE   LEADER        WORKERS                           READY
# product-team  Active  product-lead  backend-dev,frontend-dev,qa-eng.  3/3
```

--------------------------------

### Create Project Script

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/project-management/references/create-project.md

Execute this bash script to create a new project. It handles directory creation, meta.json setup, plan.md placeholder, and Matrix room creation. Ensure to fill in the full plan.md after execution.

```bash
PROJECT_ID="proj-$(date +%Y%m%d-%H%M%S)"

bash /opt/hiclaw/agent/skills/project-management/scripts/create-project.sh \
  --id "${PROJECT_ID}" \
  --title "<title>" \
  --workers "worker1,worker2,worker3"
```

--------------------------------

### Manager Statelessness: Configuration Fetching from OSS

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/design/hiclaw-controller-refactor.md

Fetching manager configuration directly from OSS in a stateless setup.

```go
配置从 OSS 拉取
```

--------------------------------

### Create Human Script Usage Examples

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/human-management/references/create-human.md

Demonstrates how to use the `create-human.sh` script for different access levels. Ensure SMTP is configured for email notifications.

```bash
bash /opt/hiclaw/agent/skills/human-management/scripts/create-human.sh \
  --matrix-id "@john:domain" --name "John Doe" --level 1 \
  --email john@example.com
```

```bash
bash /opt/hiclaw/agent/skills/human-management/scripts/create-human.sh \
  --matrix-id "@jane:domain" --name "Jane Smith" --level 2 \
  --teams alpha-team,beta-team --workers standalone-dev \
  --email jane@example.com
```

```bash
bash /opt/hiclaw/agent/skills/human-management/scripts/create-human.sh \
  --matrix-id "@bob:domain" --name "Bob" --level 3 \
  --workers alice,charlie \
  --email bob@example.com
```

--------------------------------

### Helm Chart CRD Installation

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/design/hiclaw-controller-refactor.md

Instructions for installing Custom Resource Definitions (CRDs) with the Helm chart, including Manager and DebugWorker.

```yaml
CRD 安装（含 Manager、DebugWorker）
```

--------------------------------

### Install mcporter CLI

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/copaw-worker-agent/skills/mcporter/SKILL.md

If the `mcporter` command is not found, it can be installed globally using npm. This ensures the CLI tool is available for use.

```bash
npm install -g mcporter
```

--------------------------------

### One-Click Deployment Command

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/design/team-worker-proposal.md

This command demonstrates how to deploy the company setup using the hiclaw-apply script with a specified YAML configuration file.

```bash
./hiclaw-apply.sh -f company-setup.yaml

```

--------------------------------

### Install Specific HiClaw Version

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/faq.md

Specify the desired HiClaw version using the HICLAW_VERSION environment variable during the installation script execution.

```bash
HICLAW_VERSION=v1.1.0 bash <(curl -sSL https://higress.ai/hiclaw/install.sh)
```

--------------------------------

### Build with China APT and NPM Mirrors

Source: https://github.com/agentscope-ai/hiclaw/blob/main/AGENTS.md

Use this command to specify APT and NPM mirrors for faster builds, particularly for Node.js-based images like openclaw-base.

```bash
make build-openclaw-base DOCKER_BUILD_ARGS="--build-arg APT_MIRROR=mirrors.aliyun.com --build-arg NPM_REGISTRY=https://registry.npmmirror.com/"
```

--------------------------------

### Install HiClaw Manager (macOS/Linux)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/install/README.md

Run this script to install the HiClaw Manager agent on macOS or Linux. It will prompt for configuration details.

```bash
./hiclaw-install.sh manager
# or simply
./hiclaw-install.sh
```

--------------------------------

### Setup MCP Server Script

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/mcp-server-management/references/create-update-server.md

Use this script to register a new MCP server or update an existing one with new credentials or configurations. It handles DNS registration, credential substitution, server creation/update, and configuration pushes to workers.

```bash
bash /opt/hiclaw/agent/skills/mcp-server-management/scripts/setup-mcp-server.sh \
  <server-name> <credential-value> [--yaml-file <path>] [--api-domain <domain>]
```

```bash
# Built-in template: GitHub — domain auto-extracted
bash .../setup-mcp-server.sh github "ghp_xxxxxxxxxxxx"

# User-provided YAML: custom service
bash .../setup-mcp-server.sh weather "my-key" \
    --yaml-file /tmp/mcp-weather.yaml --api-domain "api.weather.com"
```

--------------------------------

### Complete Git Workflow Request

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/worker-skills/git-delegation/SKILL.md

This example demonstrates a complete Git workflow within a single request, including cloning, configuration, branching, file creation, committing, and pushing. Plan all operations before sending.

```shell
git-request:
workspace: /root/hiclaw-fs/shared/tasks/{task-id}/workspace
operations:
  - git clone /path/to/repo.git
  - cd repo && git config user.name "alice"
  - cd repo && git config user.email "alice@hiclaw.local"
  - cd repo && git checkout -b feature/my-branch
  - mkdir -p repo/docs
  - 'cat > repo/docs/file.md << "EOF"
file content here
EOF'
  - cd repo && git add docs/file.md
  - cd repo && git commit -m "feat: add file"
  - cd repo && git push origin feature/my-branch
---CONTEXT---
Complete workflow: clone, branch, create file, commit, push
---END---
```

--------------------------------

### Local CI Simulation (Multi-arch Build)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/development.md

Perform a multi-architecture build and push, similar to how the CI pipeline operates on the main branch. Requires logging into the container registry.

```bash
# Multi-arch build like CI does on main branch
docker login ghcr.io
make push VERSION=latest REGISTRY=ghcr.io REPO=higress-group/hiclaw
```

--------------------------------

### Install HiClaw on Windows (PowerShell)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-1.0.6-release.md

Execute this PowerShell command to install HiClaw on Windows systems using PowerShell 7 or later.

```powershell
Set-ExecutionPolicy Bypass -Scope Process -Force; $wc=New-Object Net.WebClient; iex $wc.DownloadString('https://higress.ai/hiclaw/install.ps1')
```

--------------------------------

### Kine Integration for etcd Compatibility

Source: https://github.com/agentscope-ai/hiclaw/blob/main/docs/design/team-worker-proposal.md

Demonstrates how to initialize kine to provide an etcd-compatible API, using SQLite as the underlying storage backend.

```go
etcdCfg, err := endpoint.Listen(ctx, endpoint.Config{
    Listener: "127.0.0.1:2379",
    Endpoint: "sqlite:///data/hiclaw-controller/hiclaw.db",
})
```

--------------------------------

### Manually Start a Worker

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/worker-management/references/lifecycle.md

Manually starts a specified Worker. This is used when a Worker needs to be brought online outside of the auto-start mechanism.

```bash
bash /opt/hiclaw/agent/skills/worker-management/scripts/lifecycle-worker.sh --action start --worker <name>
```

--------------------------------

### Install HiClaw Worker (macOS/Linux)

Source: https://github.com/agentscope-ai/hiclaw/blob/main/install/README.md

Use this command to install a HiClaw Worker agent on macOS or Linux. Replace placeholders with your specific details.

```bash
./hiclaw-install.sh worker \
  --name alice \
  --fs http://fs-local.hiclaw.io:18080 \
  --fs-key <ACCESS_KEY> \
  --fs-secret <SECRET_KEY>
```

--------------------------------

### Install HiClaw with Non-OpenAI Provider

Source: https://github.com/agentscope-ai/hiclaw/blob/main/README.md

Install HiClaw for LLM providers that expose an OpenAI-compatible API. You must specify the `llmBaseUrl` and `defaultModel` for your provider.

```bash
helm install hiclaw higress.io/hiclaw \
  -n hiclaw-system --create-namespace \
  --render-subchart-notes \
  --set credentials.llmApiKey=<your-api-key> \
  --set credentials.llmBaseUrl=https://your-provider.example.com/v1 \
  --set credentials.defaultModel=your-model-name \
  --set credentials.adminPassword=<your-admin-password> \
  --set gateway.publicURL=http://localhost:18080
```

--------------------------------

### Manager Deploys Custom API Server

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-1.0.6-release.md

Use the setup-mcp-server.sh script to deploy the custom API server, providing the server name, access token, and the YAML configuration file.

```bash
setup-mcp-server.sh billing "eyJhbGciOiJSUzI1NiIs..." --yaml-file /tmp/mcp-billing.yaml
```

--------------------------------

### Task Directory Structure Example

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills/task-coordination/SKILL.md

Illustrates the typical file structure within a task directory, highlighting the workspace, notes, metadata, and the .processing marker file.

```text
tasks/{task-id}/
├── workspace/          # Code workspace (shared between Worker and Manager)
├── notes/              # Worker's notes, plan.md, memory (not synced by Manager)
├── meta.json           # Task metadata
└── .processing         # Processing marker file (created when work in progress)
```

--------------------------------

### Install HiClaw in Embedded Mode

Source: https://github.com/agentscope-ai/hiclaw/blob/main/blog/hiclaw-1.1.0-release.md

Use this command for a quick, single-command installation without needing a Kubernetes cluster. Ideal for individual developers.

```bash
bash -c "$(curl -fsSL https://get.hiclaw.ai)"
```

--------------------------------

### Create Qwen AI Provider

Source: https://github.com/agentscope-ai/hiclaw/blob/main/manager/agent/skills-alpha/higress-gateway-management/SKILL.md

Create a Qwen AI provider configuration. This is a native type configuration.

```bash
# Qwen (native type)
curl -X POST http://127.0.0.1:8001/v1/ai/providers \
  -b "${HIGRESS_COOKIE_FILE}" \
  -H 'Content-Type: application/json' \
  -d '{
    "type": "qwen", "name": "qwen",
    "tokens": ["<API_KEY>"], "protocol": "openai/v1",
    "tokenFailoverConfig": {"enabled": false},
    "rawConfigs": {"qwenEnableSearch": false, "qwenEnableCompatible": true, "qwenFileIds": []}
  }'
```