palyan-cli
v1.1.0
Published
Governance and operations CLI for multi-agent AI systems. Wraps Nervous System MCP (29 tools), Game MCP (4 tools), Ops MCP (8 tools), and agent dispatch into one command.
Maintainers
Readme
@palyan/cli
Governance and operations CLI for multi-agent AI systems.
palyan-cli wraps the Nervous System MCP (29 governance tools), Game MCP (4 tools), Ops MCP (8 tools), and a background agent dispatch system into one command-line interface.
Unlike MCP gateways (MintMCP, Gravitee, Strata) that govern tool calls, palyan-cli governs agent behavior - drift detection, compliance enforcement, kill switch, 7 behavioral rules mechanically enforced. Unlike coding agent CLIs (Claude Code, Codex, Aider), palyan-cli dispatches operational agents for business tasks - grant research, email outreach, legal review, accounting, content publishing.
Install
npm install -g @palyan/cliRequires Node.js 18+.
Quick Start
# Check system health (RAM, disk, CPU, processes)
palyan status
# Run full drift audit across all systems
palyan audit
# List all 9 configured agents
palyan agents
# Dispatch an agent to do background work
palyan dispatch lou "scan grants due April"
# Generate A2A Agent Cards for all family members
palyan a2a-cards --output ./cardsFull Command Reference
Layer 1 - Governance (Nervous System MCP)
| Command | Description |
|---------|-------------|
| palyan status | RAM, disk, CPU, process states, crash counts |
| palyan audit | Full drift audit across all systems |
| palyan audit drift | Drift audit with configurable scopes |
| palyan audit security | Vulnerability scan |
| palyan audit compliance | Bot standards validation |
| palyan audit pages | Public page health check |
| palyan audit chain | Tamper-proof log integrity verification |
| palyan audit deps | Dependency mapping |
| palyan audit pre-publish | Scan before npm publish |
| palyan snapshot | Full system snapshot with rollback script |
| palyan usage [days] | Token usage per bot per day |
| palyan diff | What changed since last session |
| palyan self-check | NS self-diagnosis |
| palyan kill --confirm | Emergency stop all processes |
| palyan session close | End-of-session: drift audit + propagators |
| palyan propagate | Run all 3 propagators |
| palyan fix-drift | Auto-fix doc drift |
| palyan test-deploy | 5-step test pipeline |
| palyan framework | Display NS behavioral rules |
| palyan info [topic] | NS documentation |
Layer 2 - Agent Dispatch
| Command | Description |
|---------|-------------|
| palyan agents | List all configured agents with status |
| palyan dispatch <agent> <task> | Dispatch background agent with task |
| palyan logs <agent> | Tail agent logs |
| palyan scan <lane> | Trigger 3C scanner (gov, grants, jobs) |
| palyan inbox | Read latest inbox triage |
| palyan partners | Partner tracker summary table |
| palyan deadlines | Parse deadlines from session handoff |
Layer 3 - A2A Agent Cards
| Command | Description |
|---------|-------------|
| palyan a2a-cards [--output dir] | Generate all A2A Agent Cards |
| palyan a2a-card <agent> | Show single agent card |
Cards follow the A2A Protocol v0.3.0 spec.
Configuration
The CLI reads config from (highest priority first):
- CLI flags:
--host,--port,--auth - Environment variables:
PALYAN_HOST,PALYAN_PORT,PALYAN_AUTH - RC file:
~/.palyanrc(JSON) - Defaults:
localhost:3475, no auth
Example ~/.palyanrc
{
"host": "localhost",
"nsPort": 3475,
"gamePort": 3471,
"opsPort": 3472,
"auth": ""
}Remote Server (Bridge)
For accessing a remote VPS through the bridge:
export PALYAN_BRIDGE_URL=https://api.100levelup.com/bridge/
export PALYAN_BRIDGE_SECRET=your-secret-hereA2A Compliance
Each agent card includes:
- name, description from family-roles.json
- version matching the NS package version
- url pointing to the configurable MCP endpoint
- skills mapped from each agent's specialty
- authentication scheme (bearer token)
- provider info for the Palyan Family AI System
Cards are generated per the A2A Protocol specification and can be served from .well-known/agent.json.
How It Works
palyan-cli makes JSON-RPC 2.0 calls to MCP servers running on your infrastructure:
- Nervous System MCP (port 3475) - 29 governance tools
- Game MCP (port 3471) - 4 coaching/scenario tools
- Ops MCP (port 3472) - 8 business operations tools
The agent dispatch system reads agent configurations and launches background tasks using the Claude Agent SDK.
Related
- mcp-nervous-system - The governance MCP server this CLI wraps
- A2A Protocol - Agent-to-Agent communication standard
License
MIT - Palyan Family AI System
