@testany/agent-chatter

Multi-agent conversation orchestration CLI for AI assistants.

License

UNLICENSED - Proprietary software. All rights reserved.

This software is proprietary and confidential. Unauthorized copying, distribution, modification, or use of this software is strictly prohibited.

For licensing inquiries: [email protected]

Installation

npm install -g @testany/agent-chatter

Quick Start

# Check installed AI CLI tools
agent-chatter status

# Manage registered agents
agent-chatter agents list
agent-chatter agents scan
agent-chatter agents verify

# Start interactive REPL
agent-chatter

# Generate example configuration
agent-chatter config-example

Features

• Multi-agent conversation orchestration
• Support for Claude Code, OpenAI Codex, Google Gemini CLI
• Interactive REPL with Ink-based UI
• Team configuration and management
• Session persistence and restoration

Queue Management (DROP Instruction)

When using AI teams, pending tasks can accumulate in the routing queue. The DROP instruction allows clearing outdated tasks.

Clear All Pending Tasks

When a major milestone is reached and you want to start fresh:

PRD approved. Let's move to development. [DROP: ALL] [NEXT: DevTeam]

This clears all pending questions and routes only to DevTeam.

Cancel Specific Member's Tasks

If someone's pending tasks are no longer needed:

Sarah is OOO, skip her review. [DROP: Sarah] [NEXT: Max]

This removes all of Sarah's pending tasks while keeping others.

Multiple Members

You can drop multiple members at once:

Skip design review. [DROP: Sarah, Max] [NEXT: DevLead]

Important Notes

• DROP executes before NEXT in the same message
• DROP only affects the pending queue, not the currently executing task
• Use [DROP: ALL] sparingly - it clears the entire pending queue
• This is a high-risk operation - use with care
• This instruction is for human users to type in their input; it is not automatically exposed to AI agents' system prompts

Debug Mode

Enable debug logging to troubleshoot issues:

# Using CLI flag
agent-chatter --debug

# Using environment variable
AGENT_CHATTER_DEBUG=1 agent-chatter

# Redirect debug logs to file (debug goes to stderr)
agent-chatter --debug 2>debug.log

Debug logs show Core service activities including:

• Agent verification process
• Message routing decisions
• Process spawning details

Proxy Configuration

If your network requires a proxy to access AI APIs, you can configure it in two ways:

Option 1: Environment Variables

# HTTPS proxy (recommended)
export https_proxy=http://proxy.example.com:8080

# HTTP proxy
export http_proxy=http://proxy.example.com:8080

Priority order: https_proxy > HTTPS_PROXY > http_proxy > HTTP_PROXY

Option 2: Command Line (overrides environment variables)

agent-chatter --proxy http://proxy.example.com:8080 agents verify claude

The --proxy option takes precedence over environment variables.

Proxy with Authentication

# Environment variable
export https_proxy=http://username:[email protected]:8080

# Command line
agent-chatter --proxy http://username:[email protected]:8080 agents verify claude

Security Note: Credentials in proxy URLs are automatically sanitized (removed) in all log output and error messages.

Limitations

• NO_PROXY: Not supported. All traffic goes through the configured proxy.
• SOCKS proxy: Not supported. Only HTTP/HTTPS proxies are supported.

Requirements

• Node.js >= 20.0.0
• At least one supported AI CLI installed:
  • Claude Code (claude)
  • OpenAI Codex (codex)
  • Google Gemini CLI (gemini)

Related Packages

• @testany/agent-chatter-core - Core library (MPL-2.0 OR Apache-2.0)

Support

For support and licensing: [email protected]