ExecutiveAgents

Multi-agent setup for ${COMPANY_NAME} and ${FAMILY_OFFICE_NAME} using OpenClaw.

Overview

This repository contains the configuration for 6 specialized AI agents, each with their own role, responsibilities, and security profile.

Agent Structure

${COMPANY_NAME} Agents (Business)

Agent	Role	Focus	Security
wembassy-coo-wren	Chief Operating Officer	Operations, revenue, pipeline	Full access
wembassy-cmo	Chief Marketing Officer	Lead generation, content, brand	Standard
wembassy-cro	Chief Revenue Officer	Sales, proposals, closing	Standard
wembassy-cto	Chief Technology Officer	Development, architecture, AI	Elevated
wembassy-hr	Human Resources	Team, contractors, hiring	Standard

${FAMILY_OFFICE_NAME} Agents (Private)

Agent	Role	Focus	Security
mcintosh-rm	Relationship Manager	Trust, assets, family office	Sandboxed

Configuration

agents.json

The main configuration file defining:

• Agent identities and workspaces
• Sandbox modes (off, shared, all)
• Tool permissions (allow/deny lists)
• Channel bindings

SOUL.md Files

Each agent has a SOUL.md defining:

• Role and responsibilities
• Voice and tone guidelines
• Key metrics and priorities
• Collaboration patterns
• Non-negotiable constraints

Security Profiles

Full Access (COO)

• Sandbox: off
• Tools: All allowed
• Direct access to business systems

Standard (CMO, CRO, HR)

• Sandbox: off
• Tools: Restricted (no exec/process for CMO/CRO/HR)
• Safe for business operations

Elevated (CTO)

• Sandbox: off
• Tools: All allowed (includes exec, process, apply_patch)
• Needed for development work

Sandboxed (RM)

• Sandbox: all (container per agent)
• Scope: agent (isolated)
• Tools: Significantly restricted (no exec, browser, etc.)
• Maximum security for family office

Quick Start

1. Clone this repository:

   git clone https://github.com/cmcintosh/ExecutiveAgents

2. Configure OpenClaw to use agents.json:

   openclaw config set agents.configPath ./ExecutiveAgents/agents.json

3. Start an agent:

   openclaw agent start wembassy-cmo

Agent Communication

Inter-Agent Messaging

Agents can communicate via sessions_send:

// From CMO to CRO
sessions_send({
  sessionKey: "agent:wembassy-cro",
  message: "New sales letter ready for review"
})

Human Handoff

All agents can escalate to ${CEO_FIRST_NAME} via Discord DM.

Privacy Considerations

• ${COMPANY_NAME} agents handle business operations
• mcintosh-rm agent handles sensitive family matters
• No data sharing between business and family agents
• RM agent is sandboxed with restricted tools
• Secure channels only for family office communications

Development

To add a new agent:

1. Create folder in agents/{agent-id}/
2. Write SOUL.md with role definition
3. Add entry to agents.json
4. Configure sandbox and tool permissions
5. Test in isolated environment
6. Document in this README

Documentation

Internal Documentation

• TEAM-HANDBOOK.md - Information security policies and incident reporting
• WORKFLOW.md - Agent collaboration workflow
• EMAIL_WORKFLOW.md - Email routing and forwarding protocol
• DISCORD_SETUP.md - Discord bot configuration
• SETUP.md - Deployment and server setup guide
• SKILLS_MIGRATION.md - Transfer skills to server
• PROJECTBOARD_SETUP.md - Create agent user accounts in ${PROJECTBOARD_NAME}
• DEPLOYMENT_CHECKLIST.md - Step-by-step deployment

OpenClaw Reference

• OpenClaw Multi-Agent Docs
• Sandboxing Guide
• Tool Policy
• Skills Documentation

Support

For OpenClaw agent issues:

• openclaw sandbox explain - Debug tool restrictions
• openclaw agent list - See all configured agents
• openclaw agent status {agent-id} - Check agent health

License

This project is licensed under the MIT License.