collet

Agent hub CLI — manage and run AI agents

A command-line interface for managing, monitoring, and running AI agents in a centralized hub.

Features

• 📋 List all registered agents
• 🎯 Enable/disable agents
• ▶️ Run agents with parameters
• 📊 Monitor agent status
• 📤 Export agent information (JSON, table)
• 🔧 Agent metadata and configuration

Installation

pip install collet

Quick Start

# List all agents
collet agents list

# Show hub status
collet agents status

# Run an agent
collet agents run analyzer

# Enable/disable agents
collet agents enable analyzer
collet agents disable executor

# Show agent details
collet agents info analyzer

# Export as JSON
collet agents list --format json

Commands

List Agents

# List all agents
collet agents list

# List agents by status
collet agents list --status running
collet agents list --status idle
collet agents list --status error

# Export as JSON
collet agents list --format json

# Export as table (default)
collet agents list --format table

Show Status

# Show overall hub status
collet agents status

Run Agent

# Run an agent
collet agents run analyzer

# Run with parameters
collet agents run processor --params '{"input": "data.txt"}'

Enable/Disable

# Enable an agent
collet agents enable executor

# Disable an agent
collet agents disable executor

Agent Info

# Show detailed agent information
collet agents info analyzer

Library Usage

from collet.agents import AgentHub, AgentConfig, AgentStatus

# Create hub
hub = AgentHub()

# List agents
for agent in hub.list():
    print(f"{agent.config.name}: {agent.status.value}")

# Get specific agent
agent = hub.get("analyzer")
if agent:
    print(f"Status: {agent.status.value}")
    print(f"Enabled: {agent.config.enabled}")

# Register new agent
config = AgentConfig(
    name="my-agent",
    description="My custom agent",
    version="1.0.0"
)
agent = hub.register(config)

# Change agent status
hub.set_status("my-agent", AgentStatus.RUNNING)

# Export as JSON
json_str = hub.to_json()

Default Agents

The hub comes with two default agents:

1. analyzer

   • Code analysis and insights agent
   • Capabilities: syntax-analysis, complexity-metrics

2. executor

   • Task execution and workflow agent
   • Capabilities: task-execution, workflow-orchestration

Output Examples

List Output (Table)

┏━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━┳━━━━━━━┳━━━━━━━━┓
┃ Name     ┃ Description          ┃ Version ┃ Status ┃ Enabled ┃
┡━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━╇━━━━━━━╇━━━━━━━━┩
│ analyzer │ Code analysis agent  │ 0.1.0   │ idle   │ ✓       │
│ executor │ Task execution agent │ 0.1.0   │ idle   │ ✓       │
└──────────┴──────────────────────┴─────────┴────────┴─────────┘

Total agents: 2

Status Output

Agent Hub Status
Total Agents: 2

By Status:
  Idle: 2
  Running: 0
  Error: 0
  Disabled: 0

Top Agents:
  • analyzer (idle)
  • executor (idle)

Agent States

• idle - Agent is available but not running
• running - Agent is currently executing
• error - Agent encountered an error
• disabled - Agent is disabled and cannot run

Options

--status <STATUS>     Filter by status (idle, running, error, disabled)
--format <FORMAT>     Output format (table, json) [default: table]
--params <PARAMS>     JSON parameters for agent execution
--version             Show version and exit

Environment

The CLI respects these environment variables:

• COLLET_HOME - Home directory for agent configuration (default: ~/.collet)
• COLLET_VERBOSE - Enable verbose logging (any value enables)

License

Apache-2.0

Support

For issues and questions, visit: https://github.com/collet/collet-ext